Drive failed again but now with party drive errors

[lennygman]

)I posted for support help a while back for one drive (drive 4) and the solution was to swap cables. That seemed to help but now i am getting errors on same drive and now on parity drive after i run parity check. I swapped cables to both drives but it did not seem to help. I run SMARTs test but it seem to complete without errors

(Last SMART test result: Completed without error)

 

Regardless, i wanted to swap the drives but wanted to know the right order of swapping drive (Drive 4 in my case) and Parity drive so that I will not loose any data. 

 

I already copied my data from Drive 4 to Drive 5 just in case. 

 

[0:0:0:0] disk HP v125w 1.00 /dev/sda 8.01GB

[1:0:0:0] disk ATA WDC WD20EARS-00M AB51 /dev/sdb 2.00TB

[2:0:0:0] disk ATA WDC WD30EFRX-68E 0A82 /dev/sdc 3.00TB

[3:0:0:0] disk ATA ST31500341AS CC1H /dev/sdd 1.50TB

[4:0:0:0] disk ATA WDC WD30EFRX-68E 0A82 /dev/sde 3.00TB

[5:0:0:0] disk ATA Hitachi HDS5C302 A180 /dev/sdf 2.00TB

[6:0:0:0] disk ATA WDC WD20EADS-00S 0A01 /dev/sdg 2.00TB

 

 

tower-syslog-20190128-1420.zip

tower-smart-20190128-1421 Parity Drive.zip

tower-smart-20190128-1421 - drive 4.zip

[Frank1940]

Most of the folks who help with these types of problems prefer that you would provide the (single) Diagnostics file rather than the individual files.     Tools   >>>  Diagnostics

[lennygman]

Thats right.. forgot.. added....

tower-diagnostics-20190128-1637.zip

[trurl]

We also prefer the diagnostics provided by more recent versions of Unraid. Yours is very old.

 

SMART looks OK and no disks are disabled. What exactly are you wanting to do? Cause this doesn't really make sense as it is currently worded.

2 hours ago, lennygman said:

i wanted to swap the drives but wanted to know the right order of swapping drive (Drive 4 in my case) and Parity drive so that I will not loose any data.

 

[trurl]

2 hours ago, lennygman said:

i am getting errors on same drive and now on parity drive after i run parity check.

Are these parity sync errors reported at the end of parity check? Or are you talking about numbers in the Errors column on Main?

 

2 hours ago, lennygman said:

I run SMARTs test but it seem to complete without errors

Was it an Extended test you ran?

[JorgeB]

Errors on both appear to be actual disk problems, though they might be intermittent, SMART for parity is showing warning signs, you might try replacing disk4 first and get lucky, if there are read errors on parity, try resyncing parity with the old disk4 back on the array.

[Frank1940]

Just to clarify @johnnie.black's post a bit.  He said "replace disk4".  That is to remove disk4 and set it aside and use a new disk in its place to see if you can rebuild the contents of disk4 onto a NEW disk.  Should there be a problem in reading the parity disk during the rebuild, you can put old disk 4 back into the array and see if you can build parity onto a new parity disk.  The final option would be to try to recover as many files as possible from that old disk4

[lennygman]

Thank you guys. That is the reason for my question as both my single parity and one disk showing errors and i was not clear which order to use to recover both disks. 

I actually dont care for disk 4 data as I moved it to disk 5 via Krusader ( it was empty) . So I am willing to simply remove disk 4 out of the array and replace Parity Disk with new drive and rebuild parity. 

I was just not sure how data i moved to disk 5 will show up as part of array. will UNRAID find that data/folders on disk 5?

 

Also,

Alternatively, i already moved data to disk 5 so i can replace disk 4 with a new drive and copy data back from disk 5 to disk 4 manually without rebuilding array and relying on parity drive? will that work? 

 

"Are these parity sync errors reported at the end of parity check? Or are you talking about numbers in the Errors column on Main?"

Yes i see errors on Main screen after i run by weekly Parity Check

 

"Was it an Extended test you ran?"

Yes i run extended but both showing completed without errors. a bit confusing as in logs i see read errors. 

[trurl]

37 minutes ago, lennygman said:

Alternatively, i already moved data to disk 5 so i can replace disk 4 with a new drive and copy data back from disk 5 to disk 4 manually without rebuilding array and relying on parity drive? will that work? 

Maybe not in the way you are thinking.

 

If you replace disk4 it will be rebuilt from parity, even though it is an empty filesystem. The replacement disk will not have an empty filesytem on it, or indeed any of the bits from disk4 leftover from the data that was there before.

 

Parity doesn't know anything about files, it only knows about bits. So it must be rebuilt to conform to parity.

 

The other possibility would be to New Config and rebuild parity, either with a new disk as disk4, or with no disk assigned as disk4,

 

So, if your intention was to replace disk4, there was no good reason to move the data off it. That data would have been rebuilt to the replacement. In fact, moving the data off that disk was just more work for your system when it already had some issues you were concerned with.

 

[lennygman]

3 minutes ago, trurl said:

 

 

The other possibility would be to New Config and rebuild parity, either with a new disk as disk4, or with no disk assigned as disk4,

 

1

Well copied the data to disk 5 as soon as i saw errors on disk 4. Thought was if disk 4 fails i would still have the data and I can always rebuild parity disk. And i was not sure about rebuilding just disk 4 as Parity disk also started to have read errors. I have a feel that if I just replace disk 4 with new drive and try to rebuild it, with parity disk having read errors, that would fail. 

 

If that happens, what would i do?

[trurl]

OK. I see where you were going with this.

 

Just New Config with a new disk4 and a new parity. Parity will rebuild and you can format the new disk4.

 

 

[Frank1940]

Let me make sure that I understand exact what the current status is.  Disk4 is giving problems.  But you have copied all of the files off of disk4.  The parity disk is having errors.  

 

IF  this is the current status this is what I would do.  Get a new disk to use as the parity disk.  I would run three preclear cycles (70 to 100 hrs) on that disk to make sure that it is OK and get it over the Infant Mortality failure hump.  I would remove the old disk 4 from the server and put the new parity disk in.   I would then go to   Tools    >>>  New Config   and set up a new configuration without disk4 in it.  (Probably move move the other disks all up one position.)  Assign the new disk as the parity disk.  Now start the array and let it rebuild parity.  (The less time with a potentially flaky disk in any array, the better!) 

 

(Folks may question the need for using preclear.  But I would rather have an issue with a disk before I install it than a week down the road!) 

 

If you still would like to see if those old removed disks are any good, run three cycles of preclear on them.  IF there was any problems with them during this, I would not use them in my array. 

[lennygman]

1 hour ago, Frank1940 said:

I would remove the old disk 4 from the server and put the new parity disk in.   I would then go to   Tools    >>>  New Config   and set up a new configuration without disk4 in it.  (Probably move move the other disks all up one position.)  Assign the new disk as the parity disk.  Now start the array and let it rebuild parity.  (The less time with a potentially flaky disk in any array, the better!) 

3

So I stop the array,

1. remove disk 4,
2. add new drive
3. build new configuration with remaining disks 1,2,3,5
4. keep old Parity Disk as unassigned
5. assign new disk after preclear as new Parity Disk
6. start array to rebuild party on new disk.

Question: the data and folders and shares i have now on old config and on disk 1,2,3,5 will actually be seen in new config array?? 

 

 

[trurl]

3 minutes ago, lennygman said:

Question: the data and folders and shares i have now on old config and on disk 1,2,3,5 will actually be seen in new config array??

Yes. The only thing New Config does is allow you to change your disk assignments, and optionally rebuild parity. And in your case you want to rebuild parity.

 

All data disks are accepted just as they are with their data. And none of your other settings change.

[lennygman]

So I

I stopped my array

removed disk 4 (dont really need any data on it), 

Added new Drive (sdb) that i will use as parity drive with new config

started preclear and right of the start i see the following in the logs and SMART 197?

Does these error means drive is bad?

 

 

tower-syslog-20190130-2033.zip

[lennygman]

I stopped the preclear and swapped SATA cable with brand new cable and started preclear again.. Still getting these errors. 

 

blk_update_request: I/O error, dev sdb

tower-syslog-20190130-2108.zip

[trurl]

In future please post the complete diagnostics zip and not just the syslog, as you were told at the beginning of this thread.

 

Pending sectors is a case where I think preclear is useful. You need to try to get the pending reallocated. You can either try preclearing the disk until the pending sectors go to zero or just RMA the disk.

[lennygman]

Thank you.  I was just focusing on errors for preclear not the array so did not think diag files would be needed to understand if errors is critical. 

 

Anyway, preclear just errored out with a message : 

 

Preclear: 30-01-2019 21:38

FAIL! Pre-read verification failed.

 

Drive is brand new and i changed the cable. I can RMA it but my only concern that out of total of 6 Drives (5 plus Parity) , 2 started having errors at about the same time and if that is maybe an indication of MoBo issue? 

 

 

 

tower-diagnostics-20190130-2145.zip

[JorgeB]

That disk needs to be replaced:

 

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x002f   129   001   051    Pre-fail  Always   In_the_past 845
197 Current_Pending_Sector  0x0032   200   200   000    Old_age   Always       -       47

That's not a cable related problem, it also wasn't the board, it's likely just a bad disk, though in some rare cases a bad PSU can damage disks.