Brunnhilde kernel: XFS (sdc1): metadata I/O error: block 0x16142e678 ("xfs_trans_read_buf_map") error 5 numblks 8

[wgstarks]

Sep 1 20:07:58 Brunnhilde kernel: XFS (sdc1): metadata I/O error: block 0x16142e678 ("xfs_trans_read_buf_map") error 5 numblks 8

My log is getting spammed with a constant stream of this error. Don't know if it's related, but I'm in the middle of a data rebuild after a single disk failure. Diagnostics attached.

 

brunnhilde-diagnostics-20170901-2008.zip

[JorgeB]

Old disk1 was sdc and it dropped offline and was reatached as sdm, so unRAID is confused, rebooting will clear it up.

[wgstarks]

The old disk1 was sdc but I never could get it to mount after it dropped off line so I replaced it. Not sure about the sdm, don't have one now anyway.

 

Will rebooting right now effect the data rebuild that is in progress for the new disk1? Rebuild is at about 30% right now.

[JorgeB]

18 minutes ago, wgstarks said:

Will rebooting right now effect the data rebuild that is in progress for the new disk1?

 

It will start over, you can let the rebuild finish.

[wgstarks]

Would this issue cause a data loss during rebuild? The rebuild has finished but almost all data is now missing from the disk. It was all movies that I can replace (once I figure out which ones ), so nothing critical, but I'd kinda like to know what caused the loss so it doesn't happen again.

[JorgeB]

Filesystem corruption can always cause data loss, have you rebooted since? You may need to run xfs_repair on the rebuilt disk.

[wgstarks]

Yes. I rebooted to clear the metadata I/O error.

 

If I can still run XFS repair that would be great (no idea how to do it though). If not, oh well. Just have to re-rip a bunch a bunch of videos.

[JorgeB]

xfs_repair will only help if there's fs corruption on the rebuilt disk, if you like post your current diags.

[wgstarks]

Attached

 

 

brunnhilde-diagnostics-20170903-0849.zip

[JorgeB]

No obvious fs issues, disk1 has little data but more than on your first diags, where it was practically empty, when it was rebuilding, so whatever happened it was before that.

[wgstarks]

Would a parity check (corrections enabled) effect emulated disks? Had a parity check start before I had a replacement disk available. Stopped it when I saw it was running, but that was several hours later.

[JorgeB]

No, it will act just as a read check, no data/parity will be changed, problem is more consistent with parity not being in sync when the disk failed.

[JorgeB]

8 minutes ago, johnnie.black said:

problem is more consistent with parity not being in sync when the disk failed.

 

Assuming you didn't format the disk at any point before or during the rebuild.

[wgstarks]

No. I added a new drive in an empty drive bay adjacent to disk1. Disk1 dropped out at the same time the new disk mounted. Really thought I had just jostled it and lost connection. Wasn't able to get disk1 to remount though. Would just get a message that I needed to format the disk (unmountable disk present).