In scheduler I have Write corrections to parity disk set to yes. I also have the parity check tuning plugin installed. When the scheduled parity check is paused and resumed I keep getting notifications saying it's a non correcting parity check as below. Is this just a limitation of the plugin or am I doing something wrong?

Parity Check Tuning	[UNRAID] Paused	Scheduled Non-Correcting Parity-Check (31.5% completed)

 

Edited May 3, 20242 yr by shiftylilbastrd
typo

12 hours ago, shiftylilbastrd said:

In scheduler I have Write corrections to parity disk set to yes. I also have the parity check tuning plugin installed. When the scheduled parity check is paused and resumed I keep getting notifications saying it's a non correcting parity check as below. Is this just a limitation of the plugin or am I doing something wrong?

Parity Check Tuning	[UNRAID] Paused	Scheduled Non-Correcting Parity-Check (31.5% completed)

 

 

I suspect it is almost certainly a known Unraid bug in certain Unraid releases where Unraid is not correctly starting the check and setting the check to write corrections, and the plugin is correctly reporting what is actually happening.    If you post your system's diagnostics (or at least the syslog)  we could confirm this.

 

We normally recommend that the scheduled check IS set to be non-correcting.   The rational being that you do not want hardware that is playing up to corrupt parity before you realise you have a problem.  That allows you to do some investigation after you get a check run that reports errors if you have not had something obvious that could cause errors like an unclean shutdown  Normally you would then only run a correcting check (manually) after confirming there is no hardware issue outstanding.

9 hours ago, itimpi said:

We normally recommend that the scheduled check IS set to be non-correcting.

I'll probably just disable it then. My thought process was that if a check was completed containing errors then I would just need to run it again to make the corrections. With a parity check taking over 30 hours I figured it would be a time saver but your logic does make sense.

Just now, shiftylilbastrd said:

I'll probably just disable it then. My thought process was that if a check was completed containing errors then I would just need to run it again to make the corrections. With a parity check taking over 30 hours I figured it would be a time saver but your logic does make sense.

That would be fine if you did not factor in the fact that if an array drive fails you cannot successfully rebuild without data loss if parity is not all valid.

1 hour ago, itimpi said:

if an array drive fails

There’s no way to have Unraid check to make sure all drives are present as part of the process before initiating a parity check? It already identifies if drives are missing as part of regular operations it just seems logical to incorporate that into the parity check process. 

35 minutes ago, shiftylilbastrd said:

There’s no way to have Unraid check to make sure all drives are present as part of the process before initiating a parity check? It already identifies if drives are missing as part of regular operations it just seems logical to incorporate that into the parity check process. 

Not sure of the point you are trying to make?

 

There is no standard way to do that.   However even if there was there is still the chance of a drive failing in the middle.   Unraid does not necessarily know there is any problem with a drive until actually tries to use it and then gets a failure.