Tag Archives: Promise Pegasus2 hackery

Hackery, Monitoring, Promise RAID

Using promiseutil to Find a Failing RAID Member.

Ideally, if a Promise Pegasus has a failing RAID member, or disk, we would want the Promise Utility GUI to report that. But that’s not always what happens.

When Backblaze published the SMART stats they pay attention to a few years ago, I adopted a practice of replacing drives that exhibited non-zero values for RAW_VALUES of the same SMART stats. Backblaze looks at:

  • SMART 5 – Reallocated_Sector_Count.
  • SMART 187 – Reported_Uncorrectable_Errors.
  • SMART 188 – Command_Timeout.
  • SMART 197 – Current_Pending_Sector_Count.
  • SMART 198 – Offline_Uncorrectable.

I’ve never seen SMART 187 or 188 reported by a drive member on a Promise Pegasus RAID, but the other values are there.

We can obtain the SMART status of the members of a RAID like this:

  1. In Terminal, start promiseutil
  2. At the prompt, type smart -v (with the verbose flag on).

The output will show the SMART statistics for each member of the RAID.

So, what does a failing Promise Pegasus RAID member look like?

In this example, the Promise Utility reported the health of the RAID as fine, but the performance of this RAID suggested otherwise. Here’s the top of the log:

-------------------------------------------------------
 PdId: 2
 Model Number: TOSHIBA DT01ACA2
 Drive Type: SATA
 SMART Status: Enable
 SMART Health Status: OK
 SCT Status Version:                  3
 SCT Version (vendor specific):       256 (0x0100)
 SCT Support Level:                   1
 Device State:                        SMART Off-line Data Collection executing in background (4)
 Current Temperature:                    37 Celsius
 Power Cycle Min/Max Temperature:     29/40 Celsius
 Lifetime    Min/Max Temperature:     19/44 Celsius
 Under/Over Temperature Limit Count:   0/0
 Self-test execution status:      (   0)    The previous self-test routine
                     completed without error or no self-test
                     has ever been run.
 Error logging capability:        (0x01)    Error logging supported.
 Short self-test routine 
 recommended polling time:      (   1) minutes.
 Extended self-test routine
 recommended polling time:      ( 249) minutes.
 SCT capabilities:            (0x003d) SCT Status supported.
                     SCT Feature Control supported.
                     SCT Data Table supported.
 SMART Self-test log structure revision number: 1
 No self-tests have been logged.  [To run self-tests, use: smartctl -t]
 SMART Error Log Version: 1

Hmm…no signs of trouble. We see SMART Health Status: OK, so if we were just grepping or awking for that, we’d assume that all was well. But a few lines down, we find ATA Error Count: 4. This value, doesn’t even appear on a healthy member, even with the -v verbose flag. And that’s followed by four errors.

SMART Error Log Version: 1
 ATA Error Count: 4
     CR = Command Register [HEX]
     FR = Features Register [HEX]
     SC = Sector Count Register [HEX]
     SN = Sector Number Register [HEX]
     CL = Cylinder Low Register [HEX]
     CH = Cylinder High Register [HEX]
     DH = Device/Head Register [HEX]
     DC = Device Command Register [HEX]
     ER = Error register [HEX]
     ST = Status register [HEX]
 Powered_Up_Time is measured from power on, and printed as
 DDd+hh:mm:SS.sss where DD=days, hh=hours, mm=minutes,
 SS=sec, and sss=millisec. It "wraps" after 49.710 days.
 Error 4 occurred at disk power-on lifetime: 34794 hours (1449 days + 18 hours)
   When the command that caused the error occurred,
   the device was active or idle.
 After command completion occurred, registers were:
   ER ST SC SN CL CH DH
 
 40 51 80 80 e7 1c 0d  Error: UNC 128 sectors at LBA = 0x0d1ce780 = 219998080
 Commands leading to the command that caused the error were:
   CR FR SC SN CL CH DH DC   Powered_Up_Time  Command/Feature_Name
   -- -- -- -- -- -- -- --  ----------------  --------------------
   25 00 80 80 e7 1c 40 00   2d+17:20:39.072  READ DMA EXT
   35 00 00 00 44 13 40 00   2d+17:20:39.070  WRITE DMA EXT
   35 00 00 00 35 13 40 00   2d+17:20:39.061  WRITE DMA EXT
   35 00 80 80 25 13 40 00   2d+17:20:39.052  WRITE DMA EXT
   25 00 58 a8 1f 13 40 00   2d+17:20:39.046  READ DMA EXT
 Error 3 occurred at disk power-on lifetime: 34794 hours (1449 days + 18 hours)
   When the command that caused the error occurred,
   the device was active or idle.
 After command completion occurred, registers were:
   ER ST SC SN CL CH DH
 
 40 51 78 80 b2 1e 0d  Error: UNC 120 sectors at LBA = 0x0d1eb280 = 220115584
 Commands leading to the command that caused the error were:
   CR FR SC SN CL CH DH DC   Powered_Up_Time  Command/Feature_Name
   -- -- -- -- -- -- -- --  ----------------  --------------------
   25 00 78 80 b2 1e 40 00   2d+17:20:35.539  READ DMA EXT
   35 00 00 00 ad 1e 40 00   2d+17:20:35.538  WRITE DMA EXT
   25 00 18 e8 e1 1c 40 00   2d+17:20:34.067  READ DMA EXT
   61 80 00 80 ac 1e 40 00   2d+17:20:31.037  WRITE FPDMA QUEUED
   2f 00 01 10 00 00 00 00   2d+17:20:31.036  READ LOG EXT
 Error 2 occurred at disk power-on lifetime: 34794 hours (1449 days + 18 hours)
   When the command that caused the error occurred,
   the device was active or idle.
 After command completion occurred, registers were:
   ER ST SC SN CL CH DH
 
 40 51 a8 58 e2 1c 0d
 Commands leading to the command that caused the error were:
   CR FR SC SN CL CH DH DC   Powered_Up_Time  Command/Feature_Name
   -- -- -- -- -- -- -- --  ----------------  --------------------
   61 80 28 80 ad 1e 40 00   2d+17:20:09.122  WRITE FPDMA QUEUED
   61 80 20 00 ae 1e 40 00   2d+17:20:09.121  WRITE FPDMA QUEUED
   61 80 18 80 ae 1e 40 00   2d+17:20:09.121  WRITE FPDMA QUEUED
   61 80 10 00 af 1e 40 00   2d+17:20:09.121  WRITE FPDMA QUEUED
   61 80 08 80 af 1e 40 00   2d+17:20:09.121  WRITE FPDMA QUEUED
 Error 1 occurred at disk power-on lifetime: 34706 hours (1446 days + 2 hours)
   When the command that caused the error occurred,
   the device was active or idle.
 After command completion occurred, registers were:
   ER ST SC SN CL CH DH
 
 40 51 20 78 c5 1f 0d
 Commands leading to the command that caused the error were:
   CR FR SC SN CL CH DH DC   Powered_Up_Time  Command/Feature_Name
   -- -- -- -- -- -- -- --  ----------------  --------------------
   61 80 e0 80 d8 1a 40 00      01:51:41.474  WRITE FPDMA QUEUED
   61 80 c8 00 d8 1a 40 00      01:51:41.474  WRITE FPDMA QUEUED
   61 80 c0 00 d4 1a 40 00      01:51:41.473  WRITE FPDMA QUEUED
   61 80 b8 80 d7 1a 40 00      01:51:41.473  WRITE FPDMA QUEUED
   61 80 b0 00 d7 1a 40 00      01:51:41.473  WRITE FPDMA QUEUED

All of the errors occurred on a power up of the RAID. So what do the SMART stats for Backblaze’s preferred values (bolded) look like on this drive?

SMART Attributes Data Structure revision number: 16
 Vendor Specific SMART Attributes with Thresholds:
 ID# ATTRIBUTE_NAME
     FLAG    VALUE WORST THRESH TYPE      UPDATED    WHEN_FAILED  RAW_VALUE
 1 Raw_Read_Error_Rate     
     0x000b  074   074   016    Pre-fail  Always     -            54134415
   2 Throughput_Performance  
     0x0005  139   139   054    Pre-fail  Offline    -            70
   3 Spin_Up_Time            
     0x0007  129   129   024    Pre-fail  Always     -            295 (Average 294)
   4 Start_Stop_Count        
     0x0012  097   097   000    Old_age   Always     -            15544
   5 Reallocated_Sector_Ct   
     0x0033  081   081   005    Pre-fail  Always     -            517
   7 Seek_Error_Rate         
     0x000b  100   100   067    Pre-fail  Always     -            0
   8 Seek_Time_Performance   
     0x0005  124   124   020    Pre-fail  Offline    -            33
   9 Power_On_Hours          
     0x0012  096   096   000    Old_age   Always     -            34799
  10 Spin_Retry_Count        
     0x0013  100   100   060    Pre-fail  Always     -            0
  12 Power_Cycle_Count       
     0x0032  100   100   000    Old_age   Always     -            41
 192 Power-Off_Retract_Count 
     0x0032  076   076   000    Old_age   Always     -            29286
 193 Load_Cycle_Count        
     0x0012  076   076   000    Old_age   Always     -            29286
 194 Temperature_Celsius     
     0x0002  162   162   000    Old_age   Always     -            37 (Lifetime Min/Max 19/44)
 196 Reallocated_Event_Count 
     0x0032  063   063   000    Old_age   Always     -            846
 197 Current_Pending_Sector  
     0x0022  100   100   000    Old_age   Always     -            0
 198 Offline_Uncorrectable   
     0x0008  100   100   000    Old_age   Offline    -            0
 199 UDMA_CRC_Error_Count    
     0x000a  200   200   000    Old_age   Always     -            0

As expected, no SMART 187 or 188 values. And 197 Current_Pending_Sector and 198 Offline_Uncorrectable are both 0.

But look at the RAW_VALUE for 5 Reallocated_Sector_Ct. Not good. And while it’s not on Backblaze’s list, 1 Raw_Read_Error_Rate is really high.

The other RAID elements have no errors. So we replaced the drive, and rebuilt the RAID, and performance returned to normal.

Backup, Hackery, Promise RAID

In Which the Promise Utility GUI is Not Showing Stats

You open the Promise Utility to get some stats from the Subsystem (Promise Utility > Subsystem Information), but the Promise Utility displays nothing.

You try the command line. Nada. Zip.

Fear not. Try this (your mileage may vary, entirely at your own risk, if you’re not utterly certain what you’re doing, do not do any of this):

1. Umount your Promise volume(s) first.

diskutil unmount /Volumes/NameOfVolume1

2. Unload and delete the kernel extension:

sudo kextunload -b com.promise.driver.stex
sudo rm -rf /Library/Extensions/PromiseSTEX.kext

3. Delete the Pegasus Utility:

sudo rm -rf /Applications/Promise\ Utility.app

4. Delete the Promise Utility plist:

sudo rm -rf /Users/<username>/Library/Preferences/com.promise.PromiseUtility.plist

5. Delete the LaunchDaemons plist files:

sudo rm -rf /Library/LaunchDaemons/com.promise.emaild.plist
sudo rm -rf /Library/LaunchDaemons/com.promise.bgasched.plist
sudo rm -rf /Library/LaunchDaemons/com.promise.BGPMain_R.plist
sudo rm -rf /Library/LaunchDaemons/com.promise.diskmonitor.plist

6. Delete promiseutil (The Promise Utility installer pkg will re-install this):

sudo rm -rf /usr/local/bin/promiseutil

7. Restart the Mac.

8. Install the Pegasus 6.2.9 driver and the Promise Utility.

9. See if you now have stats for the Promise Pegasus via GUI and CLI.

Hackery

Sometimes, you stumble on the right person…

…and they reveal to you a bit of magic you didn’t know existed.

In this case, it is an undocumented flag in Promise Technology’s command line utility for the Promise Pegasus2 Thunderbolt RAID, promiseutil.

As previously discussed, it appeared to be impossible to check the status of more than one Promise Pegasus enclosure from inside a script using promiseutil. We had filed a support ticket, hoping for some kind of resolution, but were told that promiseutil works as intended.

On a hunch, I reached out to someone at Promise and asked for their help investigating this issue.  I was pleasantly surprised when the contact not only took the issue seriously, he immediately looped in other support engineers to look at the problem.

After a week of back and forth about what an appropriate solution would be, perhaps a feature request, the support engineer discovered that there is an undocumented flag that allows you to specify the hba of the Promise unit you want to execute a command on.

Here’s an example. Let’s say we want to check the SMART status of two Promise Pegasus from the command line:

 promiseutil -C smart -v

will return the information for the default device only.

If you want to be explicit about which Promise Pegasus you’re checking, first get the hba numbers of the connected units:

promiseutil -C spath

The results will be something like this:

archer:~ admin$ promiseutil -C spath
=================================================
Type  #    Model        Alias   WWN          Seq
=================================================
hba   1  * Pegasus2 R4       2000-0001-5558-2fe2  1
hba   2    Pegasus2 M4       2000-0001-5558-3f92  1

Now we use the magic (apparently undocumented) -P (uppercase, not the documented lowercase) flag to specify the unit we want to look at.

promiseutil -T hba -P 1 -C smart -v

which returns the results for the first unit.

promiseutil -T hba -P 2 -C smart -v

will return results for the second unit.

My sincere thanks to the people at Promise who helped us sort this out (you know who you are) and to my fellow bug wrestler, Allen Hancock of Watchman Monitoring.

As always, be cautious with promiseutil. Its power is mighty and Bad Things® can happen if used incorrectly.

Hackery, Monitoring, Promise RAID

Scanning more than one Promise device with promiseutil

Leave a comment

So, comes the day when you have more than a single Promise Pegasus attached to a Mac and you’d like to leverage some of your utilities to check the second device.

“No problem,” you think, “I’ll just count the number of devices, then check each one in sequence.”

Except…

promiseutil is broken in one, very important way.

From inside promiseutil, the command to switch to the second unit in the chain would be something like:

spath -a chgpath -t hba -p 2

And that command works just fine. But as we’ve seen from previous work, executing promiseutil from inside a bash script requires the use of the screen command.

Executing this command from inside promiseutil run under screen does not work correctly. promiseutil appears to ignore the command and remains on the default device.

The official response from Promise is as follows:

This has been made/designed in a way to work as it is described in the KB article (and it is not a bug,but that’s how it has been designed to work) that was given on my earlier reply and it can’t used in the way that you have given and I am sorry that there are no work around available.

If you know someone at Promise and have any influence, it would be a significant improvement to have this bug removed from the next release of the promiseutil.

Heck, if you’re feeling bored, file a bug report with them here.