Question on how to deal with a red X.

[turner]

So, I don’t do a parity check nearly enough because my disks get too hot, especially during the summer. I need to change a bunch of my disks to SFSU from reiserfs so I did a parity check before hand. During the check one of my disks went off line with a red X. Unfortunately it had only gotten 16 percent done with the parity check and had corrected 1700 or so sync errors. I’m afraid that means my parity isn’t really that accurate right now. I’m curious on how borked I am now. The way I see it, I have two options. The first option is to replace the disk and rebuild, but if my parity isn’t 100% perfect, won’t I get corrupt files? My second option is to see if I can get my disk back up, and tell unbraid that my confit is fine and force it back online, redo the parity check, and hope it finishes this time, and then replace the disk and rebuild. Everything I’ve read on, states that this isn’t a good idea, so I’m not sure what the best option it to preserve my data the best I can.

 

Any advice is appreciated!

[JorgeB]

If you haven't reboot yet please post your diagnostics: Tools -> Diagnostics

 

 

[turner]

Like a dolt I powered it down without downloading the diagnostics. Lately it seems like if I can make a wrong decision with my server, I’m making it. 

[JorgeB]

Post current diags then.

[turner]

Here they are. Thanks for taking a look!

tower-diagnostics-20180215-0854.zip

[JorgeB]

SMART looks fine, the disk is on a SAS2LP, and those are known to drop disks, though without the pre-reboot diags it's more of a guess, if you were running a correcting parity check and parity was wrongly updated when the disks dropped offline, and as long as nothing was written to the emulated disk since then, it would be best to do a new config and re-sync parity, note that the array will be unprotected until done, or to play it safer, rebuild to a new disk, then do a new config with the old one.

 

You should always run non correcting parity checks, unless sync errors are expected, like after an unclean shutdown, also for the array size you should consider adding a second parity disk and getting rid of both the SASLP and the SAS2LP as both can have issues with v6.

[turner]

Thanks for the advice Johnie. I think I'll get another disk, and do the second method. I'm also going to replace my cards. It looks like I can get some HP H220s for reasonable prices and flash them. Do you think it would be best to wait to rebuild until I get the new cards? If my current ones keep dropping and end up ruining a disk, but I always hate to change too much at once.

[JorgeB]

6 hours ago, turner said:

Do you think it would be best to wait to rebuild until I get the new cards?

Only if you can get them quickly or the server is off, because it's also a risk running the array degraded for so long, especially with so many disks.