919040 Errors after parity check

[Vildkatt]

Hi.

i have my system upp and running for about 5 years. have had a few parity errors. but this time i got 900K+ it may have bean an unclean shutdown. daughter got a hold of the power button. Did one more paritycheck but still alot of errors. cant fins anything wrong with the drives and should have Write corrections to parity disk on. 

[Vildkatt]

Here is diagnostics aswell!

vildkatt-diagnostics-20221003-2014.zip

[JorgeB]

900K is a lot for an unclean shutdown, there weren't any other recent changes?

[trurl]

18 minutes ago, Vildkatt said:

should have Write corrections to parity disk on

You should NOT have that on until you determine you have parity errors and have ruled out other causes. You don't want to corrupt parity because of problems with another disk, or bad RAM, or whatever.

 

Each of those parity checks corrected different sectors.

 

Bad RAM is a very likely suspect.

 

 

[trurl]

32 minutes ago, trurl said:

Bad RAM is a very likely suspect.

You must do memtest now.

 

You shouldn't even attempt to run any computer unless RAM is working perfectly. Everything goes through RAM, your data, the OS and other executable code, everything. The cpu can't do anything with anything until it is loaded into RAM.

 

 

[Vildkatt]

Thanks. will preform memtest. nothingesle that i can think of. haven changed any harware for the last year. 

[Vildkatt]

22 hours ago, trurl said:

You must do memtest now.

 

You shouldn't even attempt to run any computer unless RAM is working perfectly. Everything goes through RAM, your data, the OS and other executable code, everything. The cpu can't do anything with anything until it is loaded into RAM.

 

 

So memtest show no errors after one pass. any other suggestions?

[JorgeB]

Run another check, if no more errors are found you are likely fine.

[Vildkatt]

27 minutes ago, JorgeB said:

Run another check, if no more errors are found you are likely fine.

started it now. finding errors from the start.

 

[Vildkatt]

1 hour ago, Vildkatt said:

started it now. finding errors from the start.

 

Doing extended smart test now. instead to se if that tells something. is there anyway to se from what disc the sync errors came from? 

[trurl]

Diagnostics taken without rebooting after each parity check are needed to compare each run. If there are obvious I/O errors that would also be apparent in diagnostics. If the only thing happening is actual sync errors, no way to determine what disk since parity is just a single bit and can only tell you that it doesn't match all the other disks combined.

[Vildkatt]

On 10/4/2022 at 11:13 PM, trurl said:

Diagnostics taken without rebooting after each parity check are needed to compare each run. If there are obvious I/O errors that would also be apparent in diagnostics. If the only thing happening is actual sync errors, no way to determine what disk since parity is just a single bit and can only tell you that it doesn't match all the other disks combined.

New day new parity check. 497622 errors. attached is diagnostics.

vildkatt-diagnostics-20221006-0750.zip

[JorgeB]

Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=1064
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=13304
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=30712
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=44016
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=58344
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=69616
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=81896
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=92136
Oct  2 11:29:43 vildkatt kernel: md: recovery thread: P corrected, sector=103400

 

Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=1064
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=13304
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=30712
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=44016
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=58344
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=69616
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=81896
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=92136
Oct  4 20:40:57 vildkatt kernel: md: recovery thread: P incorrect, sector=103400

 

Didn't check them all but at least the first ones are exactly the same sectors as before, that could indicate that they were previously wrongly updated, due to some hardware issue, run a correcting check and then a non correcting one right after.

[Vildkatt]

Did one correcting and one non correcting. 900k for the first

one and 0 for the second one. Is that acceptable pr should i make one more? 

[JorgeB]

That's good for now, of course we still don't know what caused the initial problem, use the server normally for a week or two and then run another check.