large number of signal 9 (SIGKILL) on pool www since rc 7 install

Changed Status to Open

Changed Priority to Minor

Start in safe mode and recheck

 

Is there away to do this from the CLI? GUI has become unusable, with this issue. 

Type 'reboot' in the CLI then select safe mode in the boot menu.

Should have specified I am sshing into the box. Reboot crashed with SIGKILL

If reboot didn't work try a short power key press, if it still doesn't shutdown after a few minutes you will need to force it.

Got a reboot, not a clean one but. It looks like things are behaving again. Guess I now play that whack a plugin game.

Update: So I thought I found the issue being Dynamix Cache Directories. I reinstalled everything and disabled Dynamix Cache Directories in hopes it may be patched at a later date. I woke up this morning to 100% CPU usage and 80% RAM usage I was unable to get anything out of the server. I had to reboot it via ssh again. 

I have a 100% CPU usage and 80% RAM usage again. I cant keep my box up at all on RC7 its maxing out every few hours and becoming unresponsive. I managed to get one Diag out during my multiple reboots today. 

thelibrary-diagnostics-20230607-1104.zip

Does the same thing happen in safe mode?

Yes, not as quickly. In safe mode in rc7 when I login ram is in the 60% range on rc7 and seems to steadily climb over the hours in tandem with CPU usage. In rc6 after 9 hours of usage the RAM is sitting at the 40% range and CPU usage is nominal.

With NetData installed as a Docker it reports completely different usage numbers than the UnRaid GUI with rc7 and matches the GUI in rc6.

When using Top, I see Dockerd, SFSH at the top trading places  almost exclusively in rc7 in rc6 first place is more fluid as to be expected. 

 

I had been running rc6 since almost release and have had to date no issues with it. With rc7 I haven't had more then a few hours uptime. As whatever is happening also kills all my dockers and network mounts. 

Try safe mode again but leave the docker service disabled.

OK so I had one lock up without docker or plugins, gui dead but ssh and netwrok mounts worked. 

After the reboot RAM started low but hit 80% quite quickly. Network mounts still worked. 

The control panel for ZFS was gone when I went into the drives for them.

 

thelibrary-diagnostics-20230608-1651.zip

1 hour ago, DuzAwe said:

After the reboot RAM started low but hit 80% quite quickly

That's the docker image used space, not used RAM.

Those are images I got from sperate instances of lock up. I just had another one and manged to get an image and diag out. 

I have to roll back to rc6 at this point. 

thelibrary-diagnostics-20230608-1934.zip

Your not running is safe mode, and docker is enabled.

Safe Mode no Docker

Jun  9 11:02:38 TheLibrary shfs: set -o pipefail ; /usr/sbin/zfs destroy -r 'ingress/Back Up' |& logger

 

thelibrary-diagnostics-20230609-1110.zip

That's after the mover runs, and since it's set to run hourly you can see it every hour (if there was data on that share).

So, if I don't have a crash today. Is the course of action to add one thing back at a time or ?

That's what I would recommend, first re-boot in normal mode with docker disabled, if still good re-enable docker and start one container at a time and let it run the necessary time to confirm all is good before enabling the next one.

@JorgeB Should I spin up an other thread for the missing gui for my ZFS pool?

 

Can you better describe that issue?

I'm having the same lock-ups and log errors as OP. On rc7.

 

It doesn't lock up completely just extremely unresponsive. Most dockers not working.

 

@DuzAwe, did you figure anything out with this?

@sunbear Still hunting. So far PLEX and the ARR suite has a PASS from me. I have quite a few dockers so it will be a number of days before I have gone through them all. I have also ruled out all my plugins

 

@JorgeB When I click on the first drive in the ZFS array/cluster it shows like above an incomplete interface, if I go to any other disk in the machine, Unraid Array, BTRFS and the other ZFS disks I get the normal interfaces. ie Scrub options, Smart Info, Free space setting. But for the first disk in the the ZFS it is missing all these options. 

Edited June 9, 2023 by DuzAwe

8 hours ago, DuzAwe said:

When I click on the first drive in the ZFS array/cluster it shows like above an incomplete interface

I now see the screenshot above, this was an issue with an earlier internal beta, it should not happen now, IIRC it had to do with the minimum free space calculation, @bonienlany idea?