server has been crashing non stop - at one point, accessing the plex cache folders and others in that appdata/plex/config was hanging up the system.

 

I couldnt kill that folder (was on cache drive, the nvme), but MC moved everything but those and recreated docker pulls and stood the VMs back up (moved those too). formatted the drive (set to xfs then back to brtfs) and started loading everything up again. I could swear it was just those folders in plex cache that was hanging it all up, since the other files moved fine.

 

was going well until this morning (plex was doing its mass cataloging) and it died in similar fashion

 

flash drive was recreated just in case, so fresh install. took out 2nd video card too.

 

attached diagnostic, not really sure what I am looking for.

silva-nas-diagnostics-20190318-1044.zip

Edited March 20, 20197 yr by asilva54
Solved

Why is your docker image size set to 60G? It is extremely unlikely you would ever get close to filling 20G unless you have something misconfigured. Have you been having a problem filling docker image?

no, but reducing it now since i killed the docker.img anyways to turn off auto-start on containers.

i remember seeing lots of BTRFS messages when unraid would crash on the screen i have on the side plugged into my unraid box. the HP NVME is the only new drive from a recent hardware (mobo/cpu/ram) upgrade. Taking it out now.

I don't have an NVME in my Unraid, but I do have one as the boot drive on my Windows Desktop. I had to put a heatsink on it to keep it from crashing.

I am hoping that is the case.

 

I am doing parity check now and it has sync errors, but had so many crashed that I think that is expected. 

 

I'll run a 2nd one after and if it's 0 and no crashed as I let Plex do a brand new scan, then it was the nvme

 

My logs look clean right? 

Can't really tell much from syslog since it resets when you reboot.

good point, by the time it crashed yesterday, I had no way to get in to retrieve them.

 

parity almost done. my next steps are:

• finish parity
• put in traditional sata ssd for cache (none atm, the NVMe was my cache)...hurry up amazon
• move domains/appdata/system to cache
• run new parity, pray for 0 errors

i think if i get thru all that then can safely say it was the NVMe...no crashes since taking out the NVMe.

Edited March 19, 20197 yr by asilva54

Sure enough, parity check finished. 0 errors. 

 

And that's after I put in 2 Evo 860 1TBs in a cache pool. 

 

So I don't have any real root cause, other than my Asus x470 ROG Strix didn't like the HP NVMe 1TB.