So recently I have been noticing that some of my dockers and VMs have been shutting down "by them selves".

 

All the dockers and VMs would be "up" the day before, and the next day some of them is down (red) when I check the dashboard.

 

Same thing happened this morning, and a VM was down so I started it again. But I when used the VNC browser window to access the VM, the response from the VM was extremely slow, i.e. when I typed something it would take almost 10 secs before something appeared on the VM screen.

 

So I decided a complete server reboot was maybe the best thing to do, so I tried to reboot from the Web UI. But now the Web UI has simply become unresponsive, the server doesn't reboot, and I can still access some of the dockers (so they're still still running).

 

What is the most graceful way to regain control over the server? (I would prefer not to simply "pulling the plug" on my server)

unraided-diagnostics-20170319-0632.zip

try  poweroff  from the console.

 

EDIT:     If that doesn't work, hold the power button down for a long period of time.  That will force a unclean pwerdown.

Edited March 19, 20179 yr by Frank1940

Well... the poweroff command seemed to make the server reboot after about 10 mins.

 

But still, have anyone else been experiencing "random" VM/docker container "shutdowns"?

From what I can see in your saylog (and I am not a Guru), it looks like you are having problems with memory allocation.  It looks like you have 16GB of RAM for this system and it might not be enough for everything that you are attempting to run. 

 

If someone else doesnot  get back with another idea, you might install the 'Tips and Tweaks' plugin.  That plugin has a page called "Tweaks".  I would suggest to lower the "Disk Cache 'vm.dirty_background_ratio' (%):" and the " Disk Cache 'vm.dirty_ratio (%):" settings    (I set mine to 2% and 4%.)  Turn on the Help for a description of what these settings do.  When you have a large amount of RAM installed, these settings can be lower significantly and not effect performance unless you have applications that do a large amount of file IO to a large number of files.  These settings will free up a lot of memory...

Edited March 19, 20179 yr by Frank1940

20 minutes ago, Frank1940 said:

From what I can see in your saylog (and I am not a Guru), it looks like you are having problems with memory allocation.  It looks like you have 16GB of RAM for this system and it might not be enough for everything that you are attempting to run. 

 

If someone else doesnot  get back with another idea, you might install the 'Tips and Tweaks' plugin.  That plugin has a page called "Tweaks".  I would suggest to lower the "Disk Cache 'vm.dirty_background_ratio' (%):" and the " Disk Cache 'vm.dirty_ratio (%):" settings    (I set mine to 2% and 4%.)  Turn on the Help for a description of what these settings do.  When you have a large amount of RAM installed, these settings can be lower significantly and not effect performance unless you have applications that do a large amount of file IO to a large number of files.  These settings will free up a lot of memory...

 

Thanks for the advice, I'll try it out...

 

I have previously been concerned about memory usage. I used to have a Windows VM also, but I have replaced that with an Ubuntu one to lower memory consumption.

 

Since I did that the "Memory usage", shown in the dashboard, have been bellow 50% whenever I checked it, so I'm not quite sure that the issue lies here. However, it's worth a test

 

I have also set the "Fix Common Problems" plugin to "Troubleshooting" mode, so if it happens again, then maybe I'll have some more info logged

Edited March 19, 20179 yr by Micconos

Be careful about that Memory usage stat.  One thing is that sometimes the memory request has to be contiguous and if the available memory is fragmented, the request will fail.  

 

You seem to be running Plex and I seem to recall that it is a bit of a memory hog when it builds its index.  As I look through your syslogs that are just too many 'Call Trace: ' for that not to be problems with something...