-
Notifications
You must be signed in to change notification settings - Fork 77
Drive Health and SMART
One of the most popular parts of tools for HDDs and SSDs is to pull information about how a drive is performing in order to try and predict failures.
This technology was first introduced as SMART (Self-Monitoring and Reporting Technology) in 1995 by the SSF (Small form factor) Committee.
SMART was created in order to help catch the most common causes of HDD failure of the time in a standardized way. The most common failures of drives from this time are not much different from modern drives: bad media and bad heads. There are other things that can cause pre-mature drive failure, but these are the most common.
In this Wiki page SMART and other methods of reporting drive health will be discussed as it pertains to the standards and specifications that govern modern drives and will attempt to clear up common misconceptions and incorrect interpretations of this data.
Before getting too far into all the other definitions, first SMART must be clearly understood for what it is and what it is not as well as what the standards define and what is vendor specific.
SMART on ATA disks is the most common to run across when searching the web for answers as more people interact or have interacted with ATA disks than SCSI disks.
Standardized SMART defined the following:
- Enabling or Disabling the SMART feature
- Enabling or Disabling automatically saving the SMART attributes
- Reading Attribute Threshold values (ATA-3 marks this as optional and not recommended for implementation)
- Reading Attribute Values (ATA-3 marks this as optional and not recommended for implementation)
- Return SMART Status (Pass/fail)
Later Additions to SMART Standards:
- SMART Error Logging
- SMART Offline test support (DST)
This is the original SMART command set for ATA drives. The first question many will have is
Why does ATA-3 mark the attributes and thresholds as not recommended?
This is because every drive vendor can implement their own attributes to track whatever the vendor thinks is necessary to determine drive health. There are no requirements that a specific attribute is implemented or tracked in any specific way. This is vendor specific and always has been.
The other reason this is not recommended is early on in SMART as well as in modern drives many people will attempt to compare two completely different products to each other using these values. This does not work as products and vendors may have different requirements so comparing two unequal products only leads to incorrect determination of health, reliability, quality, or leads to drawing incorrect conclusions on if a drive is healthy or not.
Attributes and their thresholds were implemented as a way for a system OEM to provide data back to the original drive vendor to assist in determining issues. Only a drive vendor will know just how the values in these attributes affect their product.
That's great, but every vendor still has attributes, and they are useful to assess a drive today. How does one use what is available in software to determine if a drive is healthy or not?
The main way in which a system is supposed to determine if a drive is healthy is the Return SMART Status command. This command simply asks the firmware if any of the attributes have dropped below their threshold value to indicate a failure....it is a simple pass/fail type of result.
More manual inspection can be done on drives that support both the attributes and the thresholds to do the same thing this command does historically.
Inside a SMART attribute there is a nominal value (also referred to as current value), a worst value, and the threshold value.
All nominal values are supposed to start at 100, as in 100% healthy and as a drive encounters errors, issues, etc. this value will drop over time. If it drops below the threshold then that attribute is considered tripped to indicate an issue.
The type of issue that a tripped attribute indicates varies based on the status flags. The standards only define one type of flag: pre-fail or advisory. This flag is in bit 0 of this field. All other flags are defined as vendor unique by the standard.
A status flag with the pre-fail/advisory flag set to 1 means that this is a pre-fail attribute. This indicates that this attribute will trip in order to predict that a drive may be failing and to begin backing up data to replace it. This is also often called a warrantied attribute since if one of these attributes causes a SMART trip during the warranty period, the vendor will usually replace the drive.
If the pre-fail/advisory status flag is set to 0 then this threshold is an advisory. An advisory is typically used to track old-age related attributes. One example would be the common Power Cycle Count. Each drive a vendor produces has an expected number of power cycles/spin up cycles that the drive is expected to be able to complete without error. Once beyond this count, the drive may continue functioning without error, however it is beyond what the vendor originally thought it would be able to provide. This does not indicate a failure or even a possibility of failure, just that the drive is beyond its expected life in respect to what this attribute is tracking.
As previously alluded to, there are some attributes that are common between drive vendors. There are no specifications that require a drive support or track any attribute, but many became common over time as system vendors set requirements for the drives they wanted to buy from different drive vendors.
This is a list of attributes that are common, but only common in name (and sometimes the attribute number). The data inside the raw values is still entirely vendor specific.
| # | Name |
|---|---|
| 1 | Read Error Rate |
| 3 | Spin Up Time |
| 4 | Start/Stop Count |
| 5 | Retired Sectors Count |
| 7 | Seek Error Rate |
| 9 | Power On Hours |
| 10 | Spin Retry Count |
| 12 | Drive Power Cycle Count |
| 187 | Reported Uncorrectable Error Count |
| 190 | Airflow Temperature |
| 194 | Temperature |
| 197 | Pending-Sparing Count or Pending Sectors Count |
| 199 | Ultra DMA CRC Error Count |
| 241 | Lifetime Writes from Host |
| 242 | Lifetime Reads from Host |
As stated above this list, these attributes are common in name and sometimes the attribute number. The Raw values a vendor reports can be different and often are different. This is not a complete list and do not expect every drive to report these attributes at these numbers...they are only common across multiple vendors.
A common example with Seagate HDDs is that the Read Error Rate raw value will output as some extremely large integer, but another vendor may have a small counter there instead.
This is because the way Seagate firmware works for this attribute is different from other vendors. It tracks multiple fields within the raw data in order to help assess the calculation of this attribute by software. Other vendors may be tracking something else in this location, hence the discrepancy and why it is important to understand that straight comparisons between vendors and SMART attributes does not make sense on its own. A high counter by one vendor for a given attribute does not have a useful meaning without knowing what it is tracking.
Another example of variation between raw values and vendors is the Temperature attribute. While it is commonly attribute 194, some vendors report only the current temperature, some report the current and highest temperature, some report a current, highest, and lowest temperature. Even when two vendors output "the same" information it may be in different byte locations or even be a different endianness (byte order). Without knowing what the vendor's format of reporting this is you cannot read this in a meaningful way.
A final example of differences between vendors that causes confusion is the lifetime reads/writes from host attributes, commonly 241 and 242. Seagate HDD's track this as a number of LBAs read or written, but Seagate SSDs have used a couple different data-units here. Some report a number of GiB read/written and others may report LBAs read/written, and another reports number of 10MB read/written. Clear as mud, isn't it?
Due to all the variation across vendors and the different attributes and their reporting schemes it has become increasingly complicated to understand what SMART attributes actually communicate to an end user.
While software like openSeaChest or smartmontools and many other exist to assist with this problem, the standards committee T13, which writes ATA drive standards, has standardized new behavior and has removed most of the original standardized SMART commands from the modern standards.
Removing the old SMART standards is not a bad thing! There is nothing in the standard telling vendors they cannot continue to support these old SMART commands, in fact many vendors will likely support them for a long time to come. The committee is simply working to reduce confusion and provide a new solution: Device Statistics.
The Device Statistics is a new log that started in the ACS-3 (ATA command set 3) standard that was created to replace the common SMART attributes with standardized statistics that report the same way, in the same place, with the same meaning various counters that drive vendors have had in SMART attributes. Looking through old committee docs there is a lot of evidence of collecting a list of what is common and what is not between vendors in order to create a common collection of data that every vendor can support in exactly the same way.
The other benefit of device statistics is that it still allows for an even greater number of statistics to be reported than is possible with SMART attributes. SMART attributes support up to 30 attributes. Device statistics currently has 53 statistics before CDL (Command Duration Limits) statistics are added with room for a vendor to implement 64 vendor specific statistics. The theoretical maximum number of statistics is 16,192 where 64 statistics are supported per page between pages 1 and 254, and still allowing 64 vendor unique statistics.
Later in this Wiki more information about device statistics and accompanying SMART attributes will be discussed about what the attribute is and what it means.
When ATA SMART was first defined, so was SMART for SCSI devices. In the T10 (SCSI) Standards committee it is called Informational Exceptions.
There are some similarities to ATA SMART, but the biggest difference is that there is no "attributes" like many are used to seeing in ATA devices. Instead, the T10 committee standardized reporting many of these same kinds of counters of various SCSI Log Pages, often times with even more detail than can be found in ATA SMART. These logs were standardized before the T10 and T13 committees (or earlier SFF committee) began working on SMART.
The first SCSI log pages to support statistics were defined in SCSI 2 (approximately 1991) and are still in use today:
- buffer over-run/under-run
- write error counters
- read error counters
- verify error counters
- read-reverse error counters
- Non-medium errors
- Last N error events
The SMART committee was mostly focused on the MRIE (Method of Reporting Informational Exceptions) field to report errors to the host.
Like ATA SMART there is a way to request the drive report its status, however it is configurable between a few different methods. This is controlled with the informational exceptions control mode page in the MRIE (Method of Reporting Informational Exceptions) field.
| MRIE | Meaning |
|---|---|
| 0 | Do not report exceptions. Similar to enabling/disabling SMART |
| 1 | Asynchronous reporting (This method is Obsolete) |
| 2 | Establish Unit Attention condition. This means sense data will report unit attention with additional sense data describing the error until an operator has intervened. |
| 3 | Conditionally Generate Recovered Error. This means the device will generate sense data with sense key "Recovered Error" and additional sense data with an error code depending on the read-write error recovery PER bit configuration. |
| 4 | Unconditionally Generate Recovered Error. Similar to more 3, but ignores the read-write error recovery PER bit. |
| 5 | Generate no sense. Sense key will be set to "No Sense" but additional sense data will indicate the error condition. Essentially this allows reporting the error without failing a command with the sense key. |
| 6 | Only report informational exception condition on request. This is most similar to how SATA works where a request sense command must be issued to determine if there is an error condition or not. |
There are numerous SCSI logs that have been created to monitor many different conditions. Some may report the exact same kinds of counters as ATA SMART attributes typically track and others may report them with more detail.
SCSI Logs and parameters:
| Log | Parameter | Data Field |
|---|---|---|
| Background Scan Results | ||
| Background Scan Status | Accumulated Power On Minutes | |
| Background Scan Status | ||
| Number of Background Scans Performed | ||
| Background Scan Progress | ||
| Number of Background Medium Scans Performed | ||
| Background Scan Results | Individual scan results reported in these parameters. | |
| Buffer Over-Run/Under-Run | ||
| Under-run | A buffer-underrun occurred. Reasons: Undefined, service delivery subsystem busy, transfer rate too slow. Experienced by: Undefined, A command, An I_T nexus, unit of time | |
| Over-run | A buffer-underrun occurred. Reasons: Undefined, service delivery subsystem busy, transfer rate too slow. Experienced by: Undefined, A command, An I_T nexus, unit of time | |
| Cache Memory Statistics | ||
| Read cache Memory Hits | ||
| Reads to Cache Memory | ||
| Write Cache Memory Hits | ||
| Writes from Cache Memory | ||
| Time from Last Hard Reset | ||
| Time Interval | ||
| Command Duration Limit Statistics | ||
| Environmental Limits | ||
| Temperature Limits | High Critical Temperature Limit Trigger | |
| High Critical Temperature Limit Reset | ||
| Low Critical Temperature Limit Reset | ||
| Low Critical Temperature Limit Trigger | ||
| High Operating Temperature Limit Trigger | ||
| High Operating Temperature Limit Reset | ||
| Low Operating Temperature Limit Reset | ||
| Low Operating Temperature Limit Trigger | ||
| Relative Humidity Limits | High Critical Relative Humidity Limit Trigger | |
| High Critical Relative Humidity Limit Reset | ||
| Low Critical Relative Humidity Limit Reset | ||
| Low Critical Relative Humidity Limit Trigger | ||
| High Operating Relative Humidity Limit Trigger | ||
| High Operating Relative Humidity Limit Reset | ||
| Low Operating Relative Humidity Limit Reset | ||
| Low Operating Relative Humidity Limit Trigger | ||
| Environmental Reporting | ||
| Temperature Report | Temperature | |
| Lifetime Maximum Temperature | ||
| Lifetime Minimum Temperature | ||
| Maximum Temperature Since Power On | ||
| Minimum Temperature Since Power On | ||
| Maximum Other Temperature | ||
| Minimum Other Temperature | ||
| Relative Humidity Report | Relative Humidity | |
| Lifetime Maximum Relative Humidity | ||
| Lifetime Minimum Relative Humidity | ||
| Maximum Relative Humidity Since Power On | ||
| Minimum Relative Humidity Since Power On | ||
| Maximum Other Relative Humidity | ||
| Minimum Other Relative Humidity | ||
| Format Status | ||
| Format Data Out | Copy of data sent in format unit command | |
| Grown Defects During Certification | ||
| Total Blocks Reassigned During Format | ||
| Total New Blocks Reassigned | ||
| Power On Minutes Since Format | ||
| General Statistics and Performance | ||
| General Access Statistics and Performance | Number of Read Commands | |
| Number of Write Commands | ||
| Number of Logical Blocks Received | ||
| Number of Logical Blocks Transmitted | ||
| Read Command Processing Intervals | ||
| Write Command Processing Intervals | ||
| Weighted Number of Read Commands + Write Commands | ||
| Weighted Read Command Processing + Write Command Processing | ||
| Idle Time | Idle Time Intervals | |
| Time Interval | Time Interval Descriptor | |
| Force Unit Access Statistics and Performance | Number of Read FUA Commands | |
| Number of Write FUA Commands | ||
| Number of Read FUA NV Commands | ||
| Number of Write FUA NV Commands | ||
| Read FUA Command Processing Intervals | ||
| Write FUA Command Processing Intervals | ||
| Read FUA NV Command Processing Intervals | ||
| Write FUA NV Command Processing Intervals | ||
| Informational Exceptions | ||
| General | Additional Sense Code | |
| Additional Sense Code Qualifier | ||
| Most Recent Temperature Reading | ||
| Non-Medium Error | ||
| Non-Medium error Count | ||
| Non-Volatile Cache | ||
| Remaining Nonvolatile Time | ||
| Maximum Nonvolatile Time | ||
| Pending Defects | ||
| Pending Defect Count | ||
| Read Error Counters | ||
| Errors Corrected without substantial Delay | ||
| Errors Corrected with possible delays | ||
| Total (e.g. rewrites or rereads) | ||
| Total Errors Corrected | ||
| Total times correction algorithm processed | ||
| Total bytes processed | ||
| Total uncorrected errors | ||
| Solid State Media | ||
| Percent Used Endurance Indicator | ||
| Start-Stop Cycle Counter | ||
| Date of Manufacture | ||
| Accounting Date | ||
| Specified Cycle Count Over Device Lifetime | ||
| Accumulated Start-Stop Cycles | ||
| Specified Load-Unload Count Over Device Lifetime | ||
| Accumulated Load-Unload Cycles | ||
| Temperature | ||
| Temperature | ||
| Reference Temperature | ||
| Utilization | ||
| Workload Utilization | ||
| Utilization Usage Rate Based On Date and Time | ||
| Verify Error Counters | ||
| Errors Corrected without substantial Delay | ||
| Errors Corrected with possible delays | ||
| Total (e.g. rewrites or rereads) | ||
| Total Errors Corrected | ||
| Total times correction algorithm processed | ||
| Total bytes processed | ||
| Total uncorrected errors | ||
| Write Error Counters | ||
| Errors Corrected without substantial Delay | ||
| Errors Corrected with possible delays | ||
| Total (e.g. rewrites or rereads) | ||
| Total Errors Corrected | ||
| Total times correction algorithm processed | ||
| Total bytes processed | ||
| Total uncorrected errors | ||
| Zoned Block Device Statistics | ||
| Maximum Open Zones | ||
| Maximum Explicitly Open Zones | ||
| Maximum Implicitly Open Zones | ||
| Minimum Empty Zones | ||
| Maximum Non-Sequential Zones | ||
| Zones Emptied | ||
| Suboptimal Write Commands | ||
| Commands Exceeding Optimal Limit | ||
| Failed Explicit Opens | ||
| Read Rule Violations | ||
| Write Rule Violations | ||
| Maximum Implicitly Open Sequential or Before Required Zones |
Additionally, a vendor may provide their own log pages (30h-3Eh) and extra parameter codes within these logs (8000h-FFFFh typically). Vendors may or may not document these additional pages in their own documentation.
Not all SCSI/SAS drives will support all of these logs, but these are a short list of logs to check how one of these drives is working. Even within the modern SAT (SCSI to ATA translation) specification, ATA Device statistics can translate directly to some of these pages.
NVMe drives support the Health log which is also called the SMART log. This log is slightly more similar to ATA SMART than SCSI SMART, but with a big difference: It is all standardized output.
Every field that is reported on this page has been part of the NVMe standard since the NVMe standards were started. There are no vendor unique counters on this log page. A vendor can still add their own logs in a vendor specific location, but this output focusses on standardization first.
This section will define various statistics that are tracked in the device statistics log as well as SMART attributes. Some vendor unique statistics/attributes may be described here, but since this software is created and maintained by Seagate, only Seagate unique data can be accurately described. Any attribute or statistics that is not described in this section may be unique to another vendor and cannot be described by anyone other than that vendor. Vendor unique statistics may change over time, so if one is missing, it may simply be new and both the software and this wiki may need updating.
The following table contains links to descriptions of various statistics/attributes to better define what they communicate about a drive.
Below is a list of SMART attributes for Seagate HDDs and some older enterprise SSDs. This list contains common attributes and some Seagate vendor unique attributes. These may link to the equivalent device statistics.
One special note: The standardized requirements in device statistics may vary slightly from a vendor's SMART attribute counter due to differences in what a vendor counts from what is required by the standard. This variation is minimal, but may be present.
All Seagate HDD's will use the attributes from this list below. There may be some variation due to what the hardware and firmware in the drive is capable of monitoring.
Examples:
- Drives with helium filled HDAs will support attribute 200, but drives with air filled HDAs will not support this attribute.
- Older generation enterprise SSDs will support 230, 231, 235 but HDDs will not support any of these.
- Older Seagate drives do not support monitoring head health with attribute 18 as this technology was developed later and is therefore only supported on newer generations.
| # | SMART Attribute Name |
|---|---|
| 1 | Read Error Rate |
| 3 | Spin Up Time |
| 4 | Start/Stop Count |
| 5 | Retired Sectors Count |
| 7 | Seek Error Rate |
| 9 | Power On Hours |
| 10 | Spin Retry Count |
| 12 | Drive Power Cycle Count |
| 18 | Head Health Self Assessment |
| 174 | Unexpected Power Loss |
| 183 | PHY Counter Events |
| 184 | IOEDC Count |
| 187 | Reported Uncorrectable Count |
| 188 | Command Timeout |
| 189 | High Fly Writes |
| 190 | Airflow Temperature |
| 191 | Shock Sensor Counter |
| 192 | Emergency Retract Count |
| 193 | Load-Unload Count |
| 194 | Temperature |
| 195 | ECC On The Fly Count |
| 197 | Pending-Sparing Count |
| 198 | Offline Uncorrectable Sector Count |
| 199 | Ultra DMA CRC Error Count |
| 200 | Pressure Measurement Limit |
| 230 | Life Curve Status |
| 231 | SSD Life Left |
| 235 | SSD Power Loss Management Life Left |
| 240 | Head Flight Hours |
| 241 | Lifetime Writes from Host |
| 242 | Lifetime Reads from Host |
| 254 | Free Fall Event Count |
Note: Many Seagate SSDs use a different attribute list depending on the model. This documentation may be revised with that information in the future.
This statistic counts the number of times a device has received a power on reset or the number on times it has uniquely been told to start up as a system is booting.
Found in:
- ATA Device statistics - general statistics page
- Common SMART attribute 12
- FARM - General drive information
- NVMe SMART/Health log - Power Cycles
Related to:
- NVMe Rotational Media Information - Spinup count
This statistic tracks the number of hours a device has been powered on. This is the time from when a power on reset has been received until the time the drive is powered off to shut a system down. This value is constantly added to between each boot cycle. Note: On an HDD this is not the same as the number of hours the spindle is spinning or the number of hours the heads are loaded above the media.
Found in:
- ATA Device statistics - general statistics page
- Common SMART attribute 9
- NOTE: Some manufacturers report in minutes.
- Seagate HDDs may also report milliseconds past the hour in some raw bytes for more granularity
- SCSI Background Scan Results log. Reports in Minutes on this page.
- NVMe SMART/Health log - this is reported by NVMe on the Health/SMART log.
- FARM - General drive information
This statistic tracks the number of logical sectors that have received a write. Multiplying this value by the drive's current logical sector size results in the number of bytes written to the drive.
Found in:
- ATA Device Statistics - general statistics page
- SMART Attribute 241
- Vendors use different units. Some report LBAs, some GiB, some GB, or other units
- SCSI Reports this in Write Error Counters log or General Statistics and Performance log
- NVMe - this is reported by NVMe on the Health/SMART log. Units are in 1000 512byte increments (a count of 1 equals 512000 bytes)
- FARM - Workload Statistics page
This statistic tracks the number of times a write to user-space sector(s) has occurred. This is a command count, which means between 1 and maximum device xfer is possible in each of these increments.
Write commands include:
- ATA Write Sectors (ext)
- ATA Write DMA (ext)
- ATA Write DMA Queued (ext)
- ATA Write FPDMA
- ATA Write Multiple (ext)
- NVMe Write
- SCSI Write (6, 10, 12, 16 byte commands)
Found in:
- ATA Device Statistics - general statistics page
- FARM - Workload Statistics page
- NVMe SMART/Health log
This statistic tracks the number of logical sectors that have received a read. Multiplying this value by the drive's current logical sector size results in the number of bytes written to the drive.
Found in:
- ATA Device Statistics - general statistics page
- SMART Attribute 242
- Vendors use different units. Some report LBAs, some GiB, some GB, or other units
- SCSI Reports this in Read Error Counters log or General Statistics and Performance log
- NVMe SMART/Health log- this is reported by NVMe on the Health/SMART log. Units are in 1000 512byte increments (a count of 1 equals 512000 bytes)
- FARM - Workload Statistics page
This statistic tracks the number of times a read to user-space sector(s) has occurred. This is a command count, which means between 1 and maximum device xfer is possible in each of these increments.
Read commands include:
- ATA Read Sectors (ext)
- ATA Read DMA (ext)
- ATA Read DMA Queued (ext)
- ATA Read FPDMA
- ATA Read Multiple (ext)
- NVMe Read
- SCSI Read (6, 10, 12, 16 byte commands)
Found in:
- ATA Device Statistics - general statistics page
- FARM - Workload Statistics page
- NVMe SMART/Health log
This statistics is used to track the current timestamp the drive has been programmed with. The statistic is set by the "Set data and time timestamp" command and uses the number of milliseconds that have elapsed since the Unix epoch (January 1st, 1970).
Found in:
- ATA Device statistics - general statistics page
- SCSI Report Timestamp command
- NVMe feature - Timestamp (set features to set, get features to read)
Pending errors are sectors the drive firmware has encountered and report errors when attempting to read them. Each count of pending errors is a unique sector in the device reporting an error.
These sectors are added to the pending defect list until a rewrite or reallocation occurs. A host can issue a write to these sectors to clear this list as the drive firmware will either rewrite it or reallocate it depending on if it can be rewritten or not. If the drive firmware reallocates the sector after the write, then the Number of Reallocated Logical Sectors counter will increase.
Found in:
- ATA Device statistics - general statistics page
- SCSI Pending Defects log
- SMART Attribute 197 or 198.
- Attribute 198 may report differently as it is often the "offline" sectors encountered. This may be errors discovered in the background during background operations. Some drives report the same count in both 197 and 198.
The workload utilization statistic is used by drive firmware to report a percentage of use. If a drive is rated for X terabytes per year workload, this statistic reports how the drive is being used relative to its rated workload. It may report a value over 100% if the drive is being used for more reads/writes than it is rated for by the vendor. Other factors that affect wear may be taken into account when calculating this value such as medium wear or head load events, etc.
Found in:
- ATA Device statistics - general statistics page
- SCSI Utilization log page
This statistic is similar to the Workload Utilization statistic however it is calculated based on the currently programmed Date and Time Timestamp statistic value. The calculated value ranges from the time of manufacture until the value programmed in the timestamp.
Found in:
- ATA Device statistics - general statistics page
- SCSI Utilization log page
This statistic outputs a percentage of resources that are available within the drive. It can be used to assess the workload of the drive. It is updated once per hour of use.
Found in:
- ATA device statistics - general statistics page
Similar to the Resource Availability statistic, but more specific to random write workloads. Once random write resources are outside of normal bounds the device's performance may degrade and the host should reduce random writes until the device is back within normal bounds.
Found in:
- ATA device statistics - general statistics page
This is a counter of the number of times the device has detected it is in free-fall (i.e. dropped). For devices with the free-fall detection they will also perform an emergency retract to prevent the heads from hitting the medium to protect the device from damage.
Found in:
- ATA device statistics - free fall statistics page
- SMART attribute 254
This counter increments whenever a shock event occurs that is greater in magnitude than the maximum rating the device. For example: After a device has entered free-fall (i.e. dropped) a overlimit shock event may occur once the device has hit the ground/some other surface. Depending on how great of a shock occurred in this event, this counter may increase.
Found in:
- ATA device statistics - free fall statistics page
- SMART attribute 191
Inside an HDD the magnetic platters are rotated using the spindle motor. This statistic tracks how many hours this motor has been in use and spinning the platters. This may be the same or less than the Power On Hours depending on when the drive has entered lower power states which may spin the drive down.
Found in:
- ATA device statistics - rotating media statistics page
- FARM - General drive information
This is the number of hours in which an HDD's heads have been flying over the magnetic medium. This includes any idle time in which the heads are still flying and not unloaded/parked. This time may differ from both the Power On Hours and Spindle Motor Power On Hours depending on which lower power states the drive has entered during its use.
Found in:
- ATA device statistics - rotating media statistics page
- SMART attribute 240 (HDD)
- FARM - General drive information
- FARM Can report separate counters for each actuator in the drive
This counter tracks the number of times an HDD has moved the heads from parked position (ramp or landing zone) to the medium. Example: From standby to active state or from idle (heads unloaded) to active state.
Found in:
- ATA device statistics - rotating media statistics page
- SMART attribute 193 (HDD)
- FARM - General drive information
- FARM can report separate counters for each actuator in the drive
- NVMe Rotational Media Information Log - Load Count
This statistic tracks the number of logical sectors that have been reallocated after device manufacture. When a sector has been marked "bad" and added to the pending list, it can be later processed during a write to determine if a rewrite can correct the sector or a reallocation is necessary to move it to an alternate location. Each time the device determines an alternate location is necessary, this statistic will be updated.
In some cases a device may support the ability to perform a reallocation during a read (Automatic Read reallocation). When this reallocation occurs, all data is preserved during the reallocation. If a device supports read reallocations, it will only be performed when the data is recovered. If the data is not recoverable, then it is lost and the device will abort the read. When the next write to that sector occurs, then the reallocation is performed and the new data with that write is in the new location and all previous data is lost.
All drives have a fixed number of sectors available for reallocation. The number of available sectors may vary by capacity or based on the vendor's design.
For drives with multiple logical sectors per physical sector (512e), this counter may grow by the number of logical sectors per physical sector. For example, a 512e drive has 8 logical sectors in a given physical sector. When the defect is found in one of those logical sectors, it affects all 8 of those sectors as they are part of the drive's physical sector with the defect. This means that the drive will most likely mark all of these bad and reallocate them all at once. This is a single physical sector reallocation, but since this counter is for logical sectors, it grew by 8. A vendor may implement the ability to perform a partial reallocation (reallocating only 1 logical sector in the physical sector), but this is up to a vendor to decide and implement.
Note:
- This statistic may report a count of logical sectors (most observed devices and SMART attributes)
- If the "normalized" bit is set to one, then this statistic reports a percentage of reallocation resources that have been used. The current/nominal SMART attribute value is scaled to a percentage similar to this normalized device statistic value.
Found in:
- ATA device statistics - rotating media statistics page
- SMART attribute 5 (HDD)
- SCSI Report defect data
- Depending on format of the defect list requested, the count may var
- Defect formats:
- Logical block
- Bytes from Offset
- Physical Cylinder-Head-Sector
- extended versions of the formats above that include more information or support larger devices
- This command can also report the primary (factory) defects as well.
- FARM - Error Statistics Page
- The FARM log can report separate counts per actuator
Read recovery attempts is a counter that records the number of logical sectors that have required 3 or more attempts to read the data from the media for each read command.
Read recovery attempts may be caused by different factors, including environmental factors. One example is using a device in a moving vehicle. The additional bumps and vibrations may cause the drive firmware to encounter an error while reading in which case it will need to try again at the next medium rotation.
Another example may be that when a device attempted to read a sector within a large server with many other drives, there was enough vibration introduced by other drives or cooling fans that required a retry at the next medium rotation to read the data from the medium without an error.
Performance may degrade when read recovery attempts are made. To reduce the impact from environmental factors be sure to use mounting designs that minimize vibration from other sources (vibration dampening mounts, mounts separate from fans and other drives, etc).
Found in:
- ATA device statistics - rotating media statistics page
- FARM - Error Statistics Page
Mechanical start failures are events that occur when an HDD is unable to start up normally. This counter tracks how many times these failures are observed. When a failure is observed by the device it may attempt a retry to correct the condition and achieve normal operation.
Found in:
- ATA device statistics - rotating media statistics page
- SMART Attribute 10
- FARM - Error Statistics Page
- NVMe Rotational Media Information Log - Failed Spin-up Count
- NVMe Rotational Media Information Log - Failed Load Count
Similar to the pending defect counter, this tracks logical sectors that are candidates for reallocation. When a pending defect has been detected the firmware has not yet decided if it needs a rewrite or reallocation. If the firmware has determined that a sector will be reallocated the next time it is written, then this statistic will increase its count.
The counter can decrement when a sector is reallocated, repairs, or a transient condition occurs.
Found in:
- ATA device statistics - rotating media statistics page
- Related to SMART Attribute 197, but not the same.
- FARM - Error Statistics Page
- FARM can report separate counters per actuator
This statistic tracks when emergency head unload events are performed. These events can be the result of:
- Processing Idle Immediate with Unload feature
- Unexpected Power loss
- device initiated self-protection (free-fall control feature)
- Another notification such as SATA Pin P11 - Direct head unload
Found in:
- ATA device statistics - rotating media statistics page
- SMART Attribute 192 (This counter may track slightly different from the requirements set above for the device statistic)
- Related to SMART Attribute 174 (Unexpected power loss)
- FARM - reliability statistics page
- Related to NVMe SMART/Health log - unsafe shutdowns
Each time the host (OS/HBA/etc) reads a sector that reports an uncorrectable error, this counter is incremented. This counter tracks each time this error is reported, it is not unique to a given sector. For example, if LBA 1000 is uncorrectable and the host issues 5 separate reads to this location, this counter will increment by 5.
Found in:
- ATA device statistics - general error statistics page
- SMART attribute 187
- NVMe SMART/Health log - Media and Data Integrity Errors. Note: This also counts CRC, ECC, checksum errors in this place so not exactly the same, but very close.
This counter increments each time a drive receives a command and either a hardware or software reset occurs before the drive provides a response from processing the command.
Another way to describe this is as a command timeout. In an operating system, when a command is issued to the drive, a time value is provided as well for the timeout value. In other words, how long does the issuer of the command expect it to take before considering it too long and to stop waiting for the completion. When this timer expires the OS may issue a hardware or software reset to the drive in order to bring it back to a state ready to accept a new command rather that continue waiting for a result of a command no longer needed.
When a timeout is set too low, this counter will increment more often.
Found in:
- ATA device statistics - general error statistics page
- SMART attribute 188
- Seagate attribute has separate fields for total count, >5 seconds, and >7.5 seconds
- FARM - Error Statistics Page
- Like the SMART attribute, separate fields for different counters
Drives that support the Storage Element Depopulation feature will support this statistic. The purpose is to increment this counter for each physical element (on an HDD, a head) has had its health go outside of the manufacturer's specified limit. In other words, the element is now reporting that it is no longer working properly.
When this happens the recommended action is to depopulate the element to continue using the rest of the drive's capacity while it is still otherwise operational.
Found in:
- ATA device statistics - general error statistics page
Related:
- SMART attribute 18 (Seagate head health self-assessment)
At the time of reading this statistics outputs the current temperature measured by the device. On SAS and SATA, this is reported in Celsius. On NVMe this is reported in Kelvin.
Found in:
- ATA device statistics - temperature statistics page
- ATA SCT status
- SMART attribute 194
- Different vendors output the reading in different locations.
- Usually found in the raw bytes but the bytes may vary
- Sometimes reported in the current/normalized value
- May be called
Primary TemperatureorHDA Temperature
- SCSI temperature log
- SCSI Environmental Reporting log
- FARM - Environmental Statistics page
- NVMe SMART/Health log (Composite temperature). NOTE: reported in Kelvin
This is an average based on the most recent 144 samples within a 24 hour period. Every 10 minutes the device enters a new reading into the list used to calculate this average. This statistic will not be valid until the first 144 samples are collected.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
This is an average temperature calculated based on the most recent 42 average short term temperature values. Every 24 hours the device will take the average short term temperature value and save it into a list used to calculate this long term average. This statistic will not be valid until the 42 readings of the average short term temperature are saved. This will take 1008 hours before this statistics is valid.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
This is the highest temperature recorded by the device since it was manufactured. This will be updated once every hour.
Found in:
- ATA device statistics - temperature statistics page
- ATA SCT status
- SMART attribute 194
- Some vendors will track this in the SMART attribute.
- Tracking may be in "worst ever" or raw bytes.
- Some vendors do not track this in SMART attributes.
- SCSI environmental reporting log
- FARM - Environmental Statistics page
This is the lowest temperature recorded by the device since it was manufactured. This will be updated once every hour.
Found in:
- ATA device statistics - temperature statistics page
- ATA SCT status
- SMART attribute 194
- Some vendors will track this in the SMART attribute.
- Tracking may be in raw bytes if this is tracked.
- Some vendors do not track this in SMART attributes.
- SCSI environmental reporting log
- FARM - Environmental Statistics page
Similar to the Average Short Term Temperature, this tracks the highest average temperature recorded.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
Similar to the Average Short Term Temperature, this tracks the lowest average temperature recorded.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
Similar to the Average Long Term Temperature, this tracks the highest average temperature recorded.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
Similar to the Average Long Term Temperature, this tracks the lowest average temperature recorded.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
This statistic tracks how many minutes the device has operated above the Specified Maximum Operating Temperature.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
- NVMe SMART/Health log - Critical Composite Temperature Time
Related to:
- ATA SCT status Over-limit count
- NVMe SMART/Health log - Warning Composite Temperature Time
This statistic specified the manufacturer's maximum temperature that the device is designed to operate in.
Found in:
- ATA device statistics - temperature statistics page
- ATA SCT status
- SCSI Temperature log (reference temperature parameter)
- SCSI Environmental Limits log
- FARM - Environmental Statistics page
This statistic tracks how many minutes the device has operated below the Specified Minimum Operating Temperature.
Found in:
- ATA device statistics - temperature statistics page
- FARM - Environmental Statistics page
Related to:
- ATA SCT status Under-limit count
This statistic specified the manufacturer's minimum temperature that the device is designed to operate in.
Found in:
- ATA device statistics - temperature statistics page
- SCSI Environmental Limits log
- FARM - Environmental Statistics page
This tracks the number of times a hardware reset has been received by the device. On SATA this includes all COMRESETS regardless of the state of the Software Settings Preservation feature.
Found in:
- ATA device statistics - transport statistics page
- FARM - General drive information
This statistics tracks how many Asynchronous Signal Recovery (ASR) events have occurred.
Another way to describe this is that the signal has been lost due to missing DWords on the interface. To minimize the impact of unsolicited COMINITs to the host, the device should only issue a COMINIT when the phy voltage falls below the minimum threshold or as the last resort during error recovery.
During recovery the device may send a COMINIT. The host will respond to this COMINIT with a COMRESET to put the device into a known state. This may reset features to their default power-on state if they are not saved to non-volatile memory.
Found in:
- ATA device statistics - transport statistics page
- FARM - Error Statistics Page
This statistic tracks each time an interface CRC error occurs since the device was manufactured. CRC errors may only be detected with protocols/interfaces that support CRCs. For example, PATA drives running PIO commands do not have CRCs associated with these commands, so it is not possible to detect a transmission error with this protocol. CRCs were introduced with UDMA protocol to minimize transmission errors as interface speeds increased.
Found in:
- ATA device statistics - transport statistics page
- SMART Attribute 199
- This attribute is also referred to as the UDMA CRC error counter due to CRCs first being introduced with UDMA protocol support.
- FARM - Error Statistics Page
On SSDs this statistic estimates the percentage of device left based on device usage and manufacturer predictions. When this statistic reports 100%, then the device is indicating all of the manufacturer's expected device life has been consumed. This does not mean the device is failing or approaching failure, only that it has reached the estimated life from the manufacturer.
A device may report values beyond 100% in this statistic.
Found in:
- ATA device statistics - solid state device statistics page
- SMART attribute 231
- The SMART attribute often is referred to as "life left". This is because it counts in the opposite direction of this statistic due to differences in how SMART and device statistics report.
- Conversion formula:
% used= 100 -life left
- SCSI Solid State Media log page
- NVMe SMART/Health log
This tracks the maximum number of zones that have been concurrently open with a zone condition of Explicitly open or Implicitly open since this statistic was last initialized.
The zones may be Sequential Write Preferred or Sequential Write Required. This statistic combines the counts for both of these zone types into this counter.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This tracks the maximum number of zones that have been concurrently open with a zone condition of Explicitly open since this statistic was last initialized.
The zones may be Sequential Write Preferred or Sequential Write Required. This statistic combines the counts for both of these zone types into this counter.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This tracks the maximum number of zones that have been concurrently open with a zone condition of Implicitly open since this statistic was last initialized.
The zones may be Sequential Write Preferred or Sequential Write Required. This statistic combines the counts for both of these zone types into this counter.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This tracks the minimum number of zones that have concurrently had a Zone Condition of Empty since this statistic was last initialized.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This is the maximum number of Sequential Write Preferred zones that have concurrently had the Non-Sequential Write Resources Active attribute set to true (nonseq = 1) since this statistic was last initialized.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This tracks the number of times the Zone Condition for any zone changed to Empty as a result of processing the Reset Write Pointer Ext command (all bit = 0).
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
Suboptimal write commands are write commands that process without an error where one or more of these conditions is present:
- the starting LBA is not at the write pointer
- The write command crosses into a write pointer where the write pointer is not at the beginning of the zone at the time it was processed
- the zone condition was FULL at the time the write command was processed
If any of these happen within a Sequential Write Preferred zone, then this statistic increments by one for each write command with one or more of these issues present.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
This statistic tracks the number of write commands that have caused a Sequential Write Preferred write pointer zone to change the zone condition to Implicitly Open and caused the number of open zones to be greater than the optimal value.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
Each time the Open Zone Ext command completes with the error Insufficient Zone Resources, this statistic increments by 1.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
Each read command that is failed with one of the following sense codes will cause this statistic to increment by 1:
- attempt to read invalid data
- read boundary violation
- zone is offline
- zone is inactive
- attempt to access gap zone
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
Each write command that is failed with one of the following sense codes will cause this statistic to increment by 1:
- insufficient zone resources
- write boundary violation
- unaligned write command
- zone is offline
- zone is read only
- zone is inactive
- attempt to access gap zone
or
- the result of attempting to write a
Sequential Write Requiredzone when the zone condition wasfull.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
The maximum number of Sequential Or Before Required zones that have been concurrently open with the zone condition of implicitly open.
Found in:
- ATA device statistics - zoned device statistics page
- SCSI Zoned Block Device Statistics
These apply to some older model enterprise SATA SSDs as well. If an attribute is not in this section then it is pointing to one of the standardized device statistics instead. This section is for SMART attributes that are unique to Seagate drives that do not have an equivalent device statistic.
The read error rate attribute is a computed value based on the number of sectors read without error and the number of sectors which had an error and required a retry or failure to read.
On a SAS drive, this can be computed using the read error counter log and verify error counter logs. TODO: Elaborate on parameters/formula.
Found in:
- SMART attribute 1
- FARM - reliability statistics page
This attribute tracks how healthy a drive's time to spin-up to ready is. The current value is a health value based on how long the drive takes to come ready past the expected time of 30 seconds.
This attribute tracks the number of retries needed to spin up the spindle in the last 8 spindle motor start attempts.
This is related to Number of Mechanical Start Failures, but not exactly the same since this counter is specific to the spindle motor.
Similar to Read Error Rate, this attribute specifically tracks seeks and seeking errors. A seek is the movement of the servo to a specific sector for a read or write operation. Whether the read or write results in an error does not affect this as a seek is only related to positioning the heads in the correct location.
Found in:
- SMART attribute 7
- FARM - reliability statistics page
Also called Head Health Self Assessment in Seagate's SMART attribute for a slightly shorter name.
The raw data of this attribute/statistic reports a bitmap of the heads within the disk. Each bit set to 1 indicates a head that has been determined as bad by the drive's firmware assessment.
This relates to the Storage Element Depopulation feature which can also report head health in the output of the Get Physical Element Status command.
This counter tracks when a power loss event occurs that is not expected. An example would be removing the power without properly shutting the system down which informs the device that power will be removed. Issuing a standby-immediate before power down will prevent this counter from incrementing.
This counter increments each time the SATA phy changes to a lower speed from what it was previously negotiated at during spin-up.
For example: A device that successfully negotiated running a 6.0Gb/s, but later is running at 3.0Gb/s will increment this counter.
IOEDC (Input/Output Error Detection Code) is a detection mechanism used for end-to-end data protection within Seagate drives. It detects when data being transmitted for read or write has encountered an error between the interface and the disk or visa-versa. Each time this detection mechanism detects an error, this counter is incremented.
The cause of an IOEDC may vary as multiple layers within the drive's hardware may hold the user data between the disk and host.
When an IOEDC error is detected, the drive will report an interface CRC error. This can allow the host to retry the command which can fix intermittent hardware failures.
Found In:
- Seagate SMART Attribute
- FARM - Error Statistics Page
A high fly write is when the head writing data to the disk is flying higher than is optimal for recording the data. This can result in a weaker magnetic field when attempting to read the sector again later. This counter increments each time this is detected.
A drive may modify the head fly height for a variety of reasons as it attempts to position the head before writing the data. Keep in mind these adjustments in height are done to navigate the peaks and valleys on the magnetic disk and may be a factor or even vibration may be a factor affecting the fly height. The magnetic heads fly within the nanometer range above the disk and these nanometer high peaks and valleys are necessary to navigate in order to get data written to the disk with a strong enough magnetic field to read it back.
This attribute reports the current temperature, highest temperature, and lowest temperature observed by a sensor on the outside of the disk, such as on the PCB. This can tell more about the operating environment and if the outside air is cooling the drive or not. Stagnant air will continue to get warmer and less effective at cooling the drive, which will cause the current temperature reading to increment.
This attribute tracks the number of time the ECC (Error Correction Code) has corrected data errors while reading data back to the host. This is related to the read error rate, but more specific to how often the ECC algorithm has corrected the data to its original state.
On a SAS drive, this is tracked as the "Total times correction algorithm processed" on the read and verify error logs.
As drive capacities have increased so has the number of discs within a hard drive. To pack more disks in and closer together Helium is used instead of air to allow closer fly heights.
Over time helium filled drives will slowly lose pressure due to helium loss. Once the pressure drops too low, these drives will no longer be able to operate without an error as the heads cannot maintain a proper fly height between the discs.
This attribute is a simple pass/fail for the drive. Once the helium pressure is too low, this attribute trips to inform the user that the drive should be replaced before any data loss occurs.
This SSD attribute helps track when a drive begins throttling. A nominal value of 90 indicates that throttling can occur depending on the drive's design requirements.
SSDs with capacitors to protect from data loss during unexpected power loss periodically test that the capacitors are working as expected. This attribute tracks the health of these capacitors. As the normalized value drops, these capacitors lose the ability to protect writes during power loss.
Outside of the requirements set by the standards committees, vendors are allowed to implement other ways of tracking drive health, performance, etc information.
This section is a short list of known public methods available.
Seagate created the FARM log, which is an acronym for Field Accessible Reliability Metrics. A public version was created and published on OCP however it has continued to advance since this time in Seagate HDD firmware.
The FARM log contains a lot of metrics that are tracked and reported in different places internal to drive firmware. In some cases the statistics in FARM match SMART attributes or device statistics as well.
As FARM has developed, more fields have been added, and some have been removed. Not all drives will report all the available fields. This depends on many factors such as the firmware revision it has, the version of FARM it supports, or if it has the necessary hardware (such as sensors) to report certain fields.
Some fields may take time to populate as the drive gets used and some fields can be populated by running Seagate's IDD (In Drive Diagnostics) test routine.
List of Subpages:
- General Drive Information
- Workload Statistics
- Error Statistics
- Environmental Statistics
- Reliability Statistics
The general drive information page of the FARM log is to provide some information about the drive the log pertains to in physical characteristics and some current configuration information.
| Field | By Head | Description |
|---|---|---|
| Model Number | The device Model Number. Matches Identify/Inquiry. NOTE: Old drives with FARM may not support this field. | |
| Serial Number | The device Serial Number. Matches what is reported in Identify/Unit Serial Number VPD page | |
| Firmware Revision | Current firmware revision of the device. Matches Identify/Inquiry | |
| World Wide Name | The device unique world wide name field. Matches identify/device identification VPD page | |
| Date of Assembly | The date the drive was assembled (week, year) | |
| Device Interface | A string describing the interface of the device. Can be SATA, SAS, or NVME
|
|
| Capacity | The capacity of the device in logical blocks. Matches identify data or read capacity data | |
| Number of LBAs (HSMR SWR capacity) | The number of LBAs on a host-managed SMR drive configured for Sequential Write Required. | |
| Physical Sector size | Size of the physical sector in bytes | |
| Logical Sector Size | Size of the logical sector in bytes | |
| Device Buffer Size | Size of the buffer/cache in bytes | |
| Number of Heads | Number of heads in the HDA | |
| Drive Recording Type | Reports if the drive is CMR or SMR or combination | |
| Form Factor | Matches the form factor reported in identify data | |
| Rotation Rate | Matches the rotation rate reported in identify data/block device characteristics VPD page | |
| ATA Security state | Copy of identify Identify word 128. SATA only. | |
| ATA Features Supported | Copy of Identify word 78. SATA only. | |
| ATA Features Enabled | Copy of Identify word 79. SATA only. | |
| Power On Hours | See Power On Hours | |
| Spindle Power On Hours | See Spindle Motor Power On Hours | |
| Head Flight hours | See Head Flying Hours NOTE: On multi-actuator drives, this is for actuator 0 | |
| Head Flight Hours, Actuator 1 | See Head Flying Hours. This reports separate values for Actuator 1 | |
| Head Load events | See Head Load Events. NOTE: On multi-actuator drives, this is for actuator 0 | |
| Head Load Events, Actuator 1 | See Head Load Events. This reports a separate counter for Actuator 1 | |
| Power Cycle Count | See Lifetime Power On Resets | |
| Hardware Reset Count | See Number of Hardware Resets | |
| Spin-Up Time | Time in milliseconds the device takes to spin up. | |
| Time to ready, last power cycle | The time a drive took to become ready in the last power cycle. Time is reported in milliseconds. | |
| Time in staggered spin-up, last power on | The time the drive was held in a staggered spin-up state waiting for the command to spin it up. Reported in milliseconds | |
| NVC Status at Power On | SAS Only. The status of the Non-volatile cache at power on | |
| Time Available to Save user Data to NVMem | The time in 100us units available by the drive to save user data to media for the non-volatile cache feature. | |
| Lowest POH timestamp | For time bounded parameters, this is the lower bound timestamp in hours | |
| Highest POH timestamp | For time bounded parameters, this is the upper bound timestamp in hours | |
| Depopulate Status | Related to Storage Element Depopulation feature. If a head has been depopulated, then this reports Depopulated otherwise it reports Not Depopulated
|
|
| Depopulated Head Mask | Related to the depopulate status. This gives a bitmask indicating which specific head(s) have been depopulated | |
| Regenerate Head Mask | Mask of heads for Regeneration | |
| Physical Element Status | ✓ | By head output of the physical element (head) status. Matches the report from Get Physical Element Status command |
| Maximum number of available disc sectors for reassignment | The maximum number of disc sectors (different from logical sectors) that are available to use for reassigning bad LBAs | |
| HAMR Data Protect Status | On HAMR drives this indicates when a drive has entered data-protect mode. In other words, write protected. | |
| POH of Most Recent FARM time series frame | The POH timestamp in the most recent time based FARM frame | |
| POH of 2nd most recent FARM time series frame | The POH timestamp in the 2nd most recent time based FARM frame. | |
| Seq or Before Req for Active Zone config | The configuration of Sequential or Before required zones on host-managed SMR drives. | |
| Seq Write Req Active Zone config | The configuration of Sequential Write Required zones on host-managed SMR drives. |
The workload statistics provide information about how a drive has been used to get an idea of what workload it has been operating under.
| Field | By Head | Description |
|---|---|---|
| Rated Workload | Percentage report based on the data the drive has collected. Note: This field is obsolete on newer drives. | |
| Total # of Read Commands | See Number of Read Commands. Note: Does not include verify commands. | |
| Total # of Write Commands | See Number of Write Commands. Note: Does not include write-verify or write same commands. | |
| Total # of Random Read Commands | The total number of random read commands to user LBA space. Note: Does not include verify commands. | |
| Total # of Random Write Commands | The total number of random write commands to user LBA space. Note: Does not include write-verify or write same commands. | |
| Total # of other commands | Total number of commands that are not reads or writes. | |
| LBAs Written | See Logical Sectors Written | |
| LBAs Read | See Logical Sectors Read | |
| # of Dither events in power cycle | Tracks the number of times dithering is performed in the current power cycle. NOTE: On dual actuator drives, this reports for actuator 0 | |
| # dither pause - random workloads in power cycle | The number of times dithering has been held off due to random workloads in the current power cycle. NOTE: On dual actuator drives, this reports for actuator 0 | |
| # dither pause - sequential workloads in power cycle | The number of times dithering has been held off due to sequential workloads in the current power cycle. NOTE: On dual actuator drives, this reports for actuator 0 | |
| # of Dither events in power cycle, Actuator 1 | Tracks the number of times dithering is performed in the current power cycle. | |
| # dither pause - random workloads in power cycle, Actuator 1 | The number of times dithering has been held off due to random workloads in the current power cycle | |
| # dither pause - sequential workloads in power cycle, Actuator 1 | The number of times dithering has been held off due to sequential workloads in the current power cycle | |
| # r cmds between 0-3.125% LBA space | Count of read commands that fall within 0-3.125% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # r cmds between 3.125-25% LBA space | Count of read commands that fall within 3.125-25% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # r cmds between 25-50% LBA space | Count of read commands that fall within 25-50% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # r cmds between 50-100% LBA space | Count of read commands that fall within 50-100% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # w cmds between 0-3.125% LBA space | Count of write commands that fall within 0-3.125% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # w cmds between 3.125-25% LBA space | Count of write commands that fall within 3.125-25% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # w cmds between 25-50% LBA space | Count of write commands that fall within 25-50% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| # w cmds between 50-100% LBA space | Count of write commands that fall within 50-100% of user LBA space. This counter is for the time reported in Time that Commands Cover (Hours) field below. |
|
| Time that Commands Cover (Hours) | This is the number of hours covered by the read/write commands in & LBA space above | |
| # r cmds with xfer <= 16KiB | Count of read commands with transfer length less than 16KiB | |
| # r cmds with xfer 16Kib - 512KiB | Count of read commands with transfer length between 16KiB and 512KiB | |
| # r cmds with xfer 512KiB - 2MiB | Count of read commands with transfer length between 512KiB and 2MiB | |
| # r cmds with xfer > 2MiB | Count of read commands with transfer length greater than 2MiB | |
| # w cmds with xfer <= 16KiB | Count of write commands with transfer length less than 16KiB | |
| # w cmds with xfer 16Kib - 512KiB | Count of write commands with transfer length between 16KiB and 512KiB | |
| # w cmds with xfer 512KiB - 2MiB | Count of write commands with transfer length between 512KiB and 2MiB | |
| # w cmds with xfer > 2MiB | Count of write commands with transfer length greater than 2MiB | |
| Time that Commands Cover (Hours) | This is the number of hours covered by the read/write xfer statistics above | |
| Queue Depth = 1 in 30s intervals | The number of times the Queue Depth was 1, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth = 2 in 30s intervals | The number of times the Queue Depth was 2, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth 3-4 in 30s intervals | The number of times the Queue Depth was 3 to 4, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth 5-8 in 30s intervals | The number of times the Queue Depth was 5 to 8, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth 9-16 in 30s intervals | The number of times the Queue Depth was 9 to 16, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth 17-32 in 30s intervals | The number of times the Queue Depth was 17 to 32, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth 33-64 in 30s intervals | The number of times the Queue Depth was 33 to 64, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Queue Depth > 64 in 30s intervals | The number of times the Queue Depth was greater than 64, recorded every 30 seconds during the time frame reported in the Time that Queue Bins Cover (Hours) field |
|
| Time that Queue Bins Cover (Hours) | This is the number of hours covered by the Queue Depth statistics above | |
| # of reads of xfer bin 4, last 3 SSF | Reads that fall into transfer length Bin 4 of the last three SMART Summary Frames | |
| # of reads of xfer bin 5, last 3 SSF | Reads that fall into transfer length Bin 5 of the last three SMART Summary Frames | |
| # of reads of xfer bin 6, last 3 SSF | Reads that fall into transfer length Bin 6 of the last three SMART Summary Frames | |
| # of reads of xfer bin 7, last 3 SSF | Reads that fall into transfer length Bin 7 of the last three SMART Summary Frames | |
| # of writes of xfer bin 4, last 3 SSF | Writes that fall into transfer length Bin 4 of the last three SMART Summary Frames | |
| # of writes of xfer bin 5, last 3 SSF | Writes that fall into transfer length Bin 5 of the last three SMART Summary Frames | |
| # of writes of xfer bin 6, last 3 SSF | Writes that fall into transfer length Bin 6 of the last three SMART Summary Frames | |
| # of writes of xfer bin 7, last 3 SSF | Writes that fall into transfer length Bin 7 of the last three SMART Summary Frames | |
| Time that XFer Bins Cover (Hours) | This is the number of hours covered by the Queue Depth statistics above |
SMART Summary Frames (SSF) are what Seagate refers to some of the internal SMART data that is tracked within the drive firmware.
Transfer length bins in SSF:
| Bin | Start | End | Start | End |
|---|---|---|---|---|
| 0 | =16KB | <16KB | ||
| 1 | >16KB | =128KB | =16KB | <512KB |
| 2 | >128KB | =512KB | =512KB | <2MB |
| 3 | >512KB | =768KB | >=2MB | |
| 4 | >768KB | =1MB | reserved | reserved |
| 5 | >1MB | =1.5MB | reserved | reserved |
| 6 | >1.5MB | =2MB | reserved | reserved |
| 7 | >2MB | reserved | reserved |
| Field | By Head | Description |
|---|---|---|
| # of Unrecoverable Read Errors | Total number of unrecoverable read command errors including errors that repeat at a given LBA location. Unrecoverable read errors are defined as read commands that are terminated with an error and report a check condition/error condition for unrecoverable media error. | |
| # of Unrecoverable Write Errors | Total number of unrecoverable write command errors including errors that repeat at a given LBA location. Unrecoverable write errors are defined as read commands that are terminated with an error and report a check condition/error condition for unrecoverable media error. | |
| # of Reallocated Sectors | See Number of Reallocated Logical Sectors. Note: Only actuator 0 | |
| # of Reallocated Sectors, Actuator 1 | See Number of Reallocated Logical Sectors. Note: Only actuator 1 | |
| # of Read Recovery Attempts | See Read Recovery Attempts | |
| # of Mechanical Start Retries | Number of retries that have been attempted to get the spindle motor spinning and up to speed. Not the same as mechanical start failures on a drive level. See Spin Retry Count. Related to Number of Mechanical Start Failures. | |
| # of Reallocation Candidate Sectors | See Number of Reallocation Candidate Logical Sectors. Note: Only actuator 0 | |
| # of Reallocation Candidate Sectors, Actuator 1 | See Number of Reallocation Candidate Logical Sectors. Note: Only actuator 1 | |
| # of ASR Events | See Number of ASR Events. NOTE: SATA Only | |
| # of Interface CRC Errors | See Number of Interface CRC Errors. NOTE: SATA Only. | |
| Spin Retry Count | SATA Only. See Spin Retry Count. Raw counter field from Seagate SMART attribute. | |
| Normalized Spin Retry Count | SATA Only. See Spin Retry Count. Nominal/Current value field from Seagate SMART attribute. | |
| Worst Ever Spin Retry Count | SATA Only. See Spin Retry Count. Worst ever field from Seagate SMART attribute. | |
| # of IOEDC Errors | See IOEDC Count. SATA Only. | |
| # of Command Timeouts | See Number of Resets Between Command Acceptance and Command Completion. SATA Only. | |
| # of Command Timeouts > 5 seconds | Related to Number of Resets Between Command Acceptance and Command Completion, but specific to > 5 seconds. Matches Seagate SMART attribute raw data field. SATA Only. | |
| # of Command Timeouts > 7.5 seconds | Related to Number of Resets Between Command Acceptance and Command Completion, but specific to > 7.5 seconds. Matches Seagate SMART attribute raw data field. SATA Only. | |
| FRU of SMART Trip Most Recent Frame | SAS Only. The Field replaceable unit code associate with the most recently logged SMART trip within the drive firmware. | |
| Port A Invalid Dword Count | SAS Only. A count of the number of interface dwords that are invalid on port A. | |
| Port B Invalid Dword Count | SAS Only. A count of the number of interface dwords that are invalid on port B | |
| Port A Disparity Error Count | SAS Only. A count of the number of times that the data encoding between the drive and the host (HBA/expander/etc) mismatch for port A. | |
| Port B Disparity Error Count | SAS Only. A count of the number of times that the data encoding between the drive and the host (HBA/expander/etc) mismatch for port B. | |
| Port A Loss of Dword Sync | SAS Only. A count of the number of times that synchronization of the bus between the drive and host (HBA/expander/etc) has been lost. When this happens the drive may be reset or attempt to resynchronize the signal. This counter is for port A. | |
| Port B Loss of Dword Sync | SAS Only. A count of the number of times that synchronization of the bus between the drive and host (HBA/expander/etc) has been lost. When this happens the drive may be reset or attempt to resynchronize the signal. This counter is for port B. | |
| Port A Phy Reset Problem | SAS Only. A counter for the number of times the port A phy has been reset due to an error condition. | |
| Port B Phy Reset Problem | SAS Only. A counter for the number of times the port B phy has been reset due to an error condition. | |
| Flash LED Info | Up to 8 Flash LED events can be reported here. Actuator 0 | |
| Flash LED Info, Actuator 1 | Up to 8 Flash LED events can be reported here for Actuator 1. | |
| Lifetime # Unrecoverable Read Errors due to ERC | Error Recovery Control feature timeout prevented further retries to read the data so the command was considered an unrecoverable read error. | |
| Cumulative Lifetime Unrecoverable Read Repeating | ✓ | Unrecoverable errors that repeat at the same sector due to retrying the read from the host. |
| Cumulative Lifetime Unrecoverable Read Unique | ✓ | Unrecoverable errors that are uniquely identified the first time the drive encounters an error at a given LBA |
| SMART Trip Flags 1 | A bit field representing different SMART trips that occurred. | |
| SMART Trip Flags 2 | A bit field representing different SMART trips that occurred. | |
| # Reallocated Sectors since last FARM Time Series Frame | The number of sectors that have been reallocated since the last time a FARM time-series snapshot was taken | |
| # Reallocated Sectors between N & N-1 FARM Time Series Frame | The number of sectors that were reallocated between the last two FARM time series frames. | |
| # Reallocation Candidate Sectors since last FARM Time Series Frame | The number of sectors that are candidates for reallocation since the last FARM time series frame was saved. | |
| # Reallocation Candidate between N & N-1 FARM Time Series Frame | The number of sectors that are candidates for reallocation since the last two FARM time series frames. | |
| # Reallocated Sectors since last FARM Time Series Frame, Actuator 1 | The number of sectors that have been reallocated since the last time a FARM time-series snapshot was taken | |
| # Reallocated Sectors between N & N-1 FARM Time Series Frame, Actuator 1 | The number of sectors that were reallocated between the last two FARM time series frames. Actuator 1. | |
| # Reallocation Candidate Sectors since last FARM Time Series Frame, Actuator 1 | The number of sectors that are candidates for reallocation since the last FARM time series frame was saved. Actuator 1. | |
| # Reallocation Candidate between N & N-1 FARM Time Series Frame, Actuator 1 | The number of sectors that are candidates for reallocation since the last two FARM time series frames. Actuator 1. | |
| # Unique Unrecoverable sectors since last FARM Time Series Frame | ✓ | Number of uniquely identified unrecoverable sectors since the last FARM time series frame was saved. |
| # Unique Unrecoverable sectors between N & N-1 FARM Time Series Frame | ✓ | Number of uniquely identified unrecoverable sectors between the last two FARM time series frames were saved. |
Flash LED: This term means an error in the firmware occurred and set an error code. These are considered severe errors. The term Flash LED is a holdover from a when error codes would be displayed by an LED on the drive blinking a specific pattern. While this LED no longer exists, the term remains.
| Field | By Head | Description |
|---|---|---|
| Current Temperature (C) | See Current Temperature | |
| Highest Temperature (C) | See Highest Temperature | |
| Lowest Temperature (C) | See Lowest Temperature | |
| Average Short Term Temperature (C) | See Average Short Term Temperature | |
| Average Long Term Temperature (C) | See Average Long Term Temperature | |
| Highest Average Short Term Temperature (C) | See Highest Average Short Term Temperature | |
| Lowest Average Short Term Temperature (C) | See Lowest Average Short Term Temperature | |
| Highest Average Long Term Temperature (C) | See Highest Average Long Term Temperature | |
| Lowest Average Long Term Temperature (C) | See Lowest Average Long Term Temperature | |
| Time in Over Temperature (Hours) | See Time in Over Temperature | |
| Time in Under Temperature (Hours) | See Time in Under Temperature | |
| Specified Max Temperature (C) | See Specified Maximum Operating Temperature | |
| Specified Min Temperature (C) | See Specified Minimum Operating Temperature | |
| Current Relative Humidity (%) | The current relative humidity as a percentage from 0-100% within the head disc enclosure. | |
| Current Motor Power Scalar | The current power scalar value used by the servo to keep the motor spinning | |
| Time Coverage for Motor Power (Hours) | The number of hours that covers the Current Motor Power Statistic value reading | |
| Current 12v input (V) | Current voltage on the 12V power line | |
| Min 12v input (V) | Minimum voltage observed on the 12V power line | |
| Max 12v input (V) | Maximum voltage observed on the 12V power line | |
| Current 5v input (V) | Current voltage on the 5V power line | |
| Min 5v input (V) | Minimum voltage observed on the 5V power line | |
| Max 5v input (V) | Maximum voltage observed on the 5V power line | |
| Time Coverage for 12v & 5v voltage (Hours) | The number of hours that covers the 12v and 5v voltage readings above. | |
| Average 12v power (W) | Average power in Watts on the 12V power line | |
| Min 12v power (W) | Minimum power in Watts on the 12V power line | |
| Max 12v power (W) | Maximum power in Watts on the 12V power line | |
| Average 5v power (W) | Average power in Watts on the 5V power line | |
| Min 5v power (W) | Minimum power in Watts on the 5V power line | |
| Max 5v power (W) | Maximum power in Watts on the 5V power line | |
| Time Coverage for 12v & 5v power (Hours) | The number of hours that covers the 12v and 5v power readings above. |
Many fields in this section report their values for each head separately. The output will use the field name + head number in the openSeaChest output.
In this section of FARM, this reports how different Seagate technologies inside the drive are functioning. These technologies can affect data integrity, and error recovery capabilities.
| Technology | Definition |
|---|---|
| DOS | Directed Offline Scan |
| ISP | Intermediate Super Parity |
| DVGA | Delta Variable Gain Amplifier |
| RVGA | Running Average Variable Gain Amplifier |
| FVGA | Filter Variable Gain Amplifier |
| MR Head Resistance | Magneto Resistive head resistance |
| H2SAT | A Head Self-Assessment Test |
| Super Parity | A Seagate reliability feature designed to improve unrecoverable read error rate |
| Velocity Observer | The divergence of the actuator coil requested current to the measured current during seek operation |
| RAW | Read After Write. Related to High Fly Writes. |
| Field | By Head | Description |
|---|---|---|
| # DOS Scans Performed | Total number of DOS Scans performed. Actuator 0. | |
| # LBAs corrected by ISP | Total number of Logical Sectors corrected due to ISP. Actuator 1 | |
| # DOS Scans Performed Actuator 1 | Total number of DOS Scans performed. Actuator 1. | |
| # LBAs corrected by ISP Actuator 1 | Total number of Logical Sectors corrected due to ISP. Actuator 1. | |
| DVGA Skip Write Detect | ✓ | The number of times a write operation was stopped due to readings from DVGA. |
| RVGA Skip Write Detect | ✓ | The number of times a write operation was stopped due to readings from RVGA. |
| FVGA Skip Write Detect | ✓ | The number of times a write operation was stopped due to readings from FVGA. |
| Skip Write Detect Threshold Exceeded | ✓ | The number of times a write operation was stopped due to a servo sample indicating a higher fly height than expected. Related to High Fly Writes attribute. |
| # Read After Write (RAW) Operations | SAS Only. The number of times the firmware has detected the need to read a sector after it has been written. | |
| Read Error Rate | SATA Only. See Read Error Rate | |
| Read Error Rate Normalized | SATA Only. See Read Error Rate. Normalized/current value reported same as in SMART | |
| Read Error Rate Worst Ever | SATA Only. See Read Error Rate. Worst ever reported same as in SMART. | |
| Seek Error Rate | SATA Only. See Seek Error Rate | |
| Seek Error Rate Normalized | SATA Only. See Seek Error Rate. Normalized/current value reported same as in SMART | |
| Seek Error Rate Worst Ever | SATA Only. See Seek Error Rate. Worst ever reported same as in SMART. | |
| High Priority Unload Events | See Number of High Priority Unload Events | |
| MR Head Resistance | ✓ | Old drives report ohms, new drives report % change since manufacturing. This is used to track changes in head performance over time. |
| Second MR Head Resistance | ✓ | Old drives report ohms, new drives report % change since manufacturing. This is used to track changes in head performance over time. |
| # of Velocity Observer | ✓ | Tracks the number of servo velocity observer errors that have been detected during seek mode operations in the last 3 SSFs. The velocity observer is the difference between demanded and delivered seek current. As this count increases, it can lead to soft seek errors. |
| # of Velocity Observer No TMD | ✓ | Tracks the number of servo timing mark detects that have been missed during seek mode operations in the last 3 SSFs. As this count increases, it can lead to soft seek errors. |
| Time Coverage for Velocity Observer (Hours) | The time in hours that covers the values in the velocity observer statistics above. | |
| H2SAT Trimmed Mean Bits in Error (Z1, Z2, Z3) | ✓ | Tracks the number of mean bits in error as calculated from the read channel in processed codewords. The data originated from a dedicated measurement on a non-user test track. |
| H2SAT Iterations to Converge (Z1, Z2, Z3) | ✓ | Tracks the number of software retries done in between read retries during error recovery process. This is an "iteration" to converge during measurement on an non-user test track. |
| Average H2SAT % Codeword at Iteration Level | ✓ | Tracks the percentage of codewords that have converged at the specified iteration to converge during the test measurement. |
| Average H2SAT Amplitude | ✓ | Measures the amplitude from the read channel as compensated by VGA response from a previous read operation. |
| Average H2SAT Asymmetry | ✓ | Measures the asymmetry value from the read channel as compensated by the VGA response from a previous read operation. |
| FAFH Appd Clr Delta (1/1000 A) (OD, ID, MD) | ✓ | Applied Fly Height Clearance delta monitors changes in head fly height at the outer, inner, and middle diameter of the disk over time. Measured in 1/1000 of an angstrom. 1 angstrom = 0.1 nanometers. 1/1000 angstrom = 0.1 picometers |
| # Disc Slip Recalibrations Performed | When an HDD is mishandled (ex: dropped) it may cause the magnetic disc to shift position. Servo firmware can correct for disc slip, but only so much before it can no longer adjust for being off of the expected position. This counts how many times this recalibration has run to adjust for changes in the disc position. | |
| # Reallocated Sectors | ✓ | See Reallocated Sectors. NOTE: This is reported per head versus for the whole drive. This is reported as a number of disc sectors. |
| # Reallocated Candidate Sectors | ✓ | See Number of Reallocation Candidate Logical Sectors NOTE: This is reported per head versus for the whole drive. This is reported as a number of disc sectors. |
| Helium Pressure Threshold | See Helium Pressure Limit Reached | |
| # DOS Ought To Scan | ✓ | The number of times DOS is recommended to scan an area of the medium. These scans are only done during idle operations. |
| # DOS Need To Scan | ✓ | The number of times DOS has been marked as needing to scan an area of the medium. These scans are done during idle operations. |
| # DOS Write Fault Scans | ✓ | The number of times DOS has been run due to write faults detected. These scans occur when a write fault exceeds an unsafe limit that can put adjacent tracks at risk. |
| Write Workload Power-on Time (Hours) | ✓ | The number of hours that the drive has had a write workload. |
| # LBAs Corrected By Parity Sector | A count of the number of LBAs that were corrected by use of a parity sector. On multi-actuator drives, this count is for actuator 0. | |
| # LBAs Corrected By Parity Sector Actuator 1 | A count of the number of LBAs that were corrected by use of a parity sector. On multi-actuator drives, this count is for actuator 1. | |
| Primary Super Parity Coverage % | Super Parity coverage for the drive reported as a percentage. | |
| Primary Super Parity Coverage SMR/HSMR-SWR % | Super Parity coverage for SMR/HSMR drives with zones configured as Sequential Write Required (SWR) | |
| Primary Super Parity Coverage SMR/HSMR-SWR % Actuator 1 | Super Parity coverage for SMR/HSMR drives with zones configured as Sequential Write Required (SWR) on Actuator 1 |
The open compute project is a collection of vendors and their data center customers. They work together on making new technologies that can better help meet the needs of a data center through hardware, firmware, and software requirements and specifications.
The project has created a specification on features and drive logs to report more information about Data Center NVMe SSDs. This log has a lot of additional information in it that is not required by the standards and may be available on some drives.
Future expansion: Add all details of OCP NVMe SSD logs/features here.