Something caused my server to restart this morning and I woke up to find two drives with the Unmountable: No file system statuses. Here is where I messed up big time. I stopped the array to restart in maintenance mode to run a smart check of the drives and go from there but accidentally started in regular mode which initiated a parity rebuild...

 

I have a found a few threads with this same issue and they all point towards the filesystem needing to be repaired. Assuming I already have not jacked everything up, will letting the parity rebuild finish overwite data on those drives or corrupt the parity data? Should I stop it? It also seems to have an unusually long estimated time at approximately 6 days vs. the normal 1 day. I attached logs and a screenshot.

 

Both drives are on the same power / sata hotswap bay so I think the original issue may be related to that because it seems weird for two drives to fail at the same time without something common between them.

mediavault-diagnostics-20200526-1054.zip

11 minutes ago, Civrock said:

Should I stop it?

The damage (if any) is already done, you'll need to run xfs_repair once it finishes, you can also cancel and run xfs_repair now then start rebuilding again from the start, this way you know if it's worth rebuilding and can also access the data during it.

 

12 minutes ago, Civrock said:

It also seems to have an unusually long estimated time at approximately 6 days vs. the normal 1 day.

There's something else using the array, in the screenshot you can see extra reads on disk6, and on the diags there extra reads on disk3.

19 minutes ago, Civrock said:

Should I stop it?

Yes, I will unassign all other disk ( data and parity ), then retain those two disks. If repair filesystem not success, than resume all assignment and rebuild.

 

19 minutes ago, Civrock said:

estimated time at approximately 6 days

Extra reading on disk6, as you see the read was 91.1 MB/s .

Edited May 26, 20206 yr by Benson

Thanks for your time guys, I will just go ahead and stop it now and attempt to repair the filesystems. Will update if I have more issues or make it through that.

Edited May 26, 20206 yr by Civrock

Okay, so I ran xfs_repiar -v on both disks and they both had issues with the primary crc, said that there were pending changes to the log and I should mount first before re-running the repair tool. When I mounted them, both showed up normally and rebuild started.

 

The wording of the repair report made it seem like (at least to me as a layperson in this area) I needed to mount and then immediately run the repair again. Since both disks showed up normally, should I just let the rebuild run, or do I need to stop and run the repair again? I feel like the right answer here is to let the rebuild run now, but I just want to be sure.

 

Thanks again for your time.

If they are mounting correctly they should be fine but it's also OK to run xfs_repair again to make sure.