hey all. i recently upgraded my CPU, mobo, ram, and SATA card(s) (basically almost everything aside from PSU and HDDs) from an old AMD Phenom II based system, with a jumble of random RAID/SATA controllers. It was working fine and was stable with the old hardware, but just felt like it was time for an upgrade.

 

Every couple of days I'm seeing issues where CPU utilization will spike up and load averages are hitting high double digits (20-40+). Most things I'm running (dockers) still work, albeit very slow. Usually I'm able to resolve it by starting/stopping some of the dockers at random. Eventually I'll just stop/start all of them and that almost always fixes it.

 

The last time this happened, I ran docker stats first and none of them were particularly using a lot of CPU, one of them was at 25% and the rest were at single digits. Restarting that docker resolved the issue, leading me to believe that is where the issue is.

 

I haven't changed or added any dockers since changing hardware. The new hardware should be noticeably better than my old hardware but it seems like it's struggling to keep up with the load at certain times, whereas on the old hardware I've gone months and month without touching it at all. Any thoughts/suggestions?

 

Current hardware:

Gigabye AB-350 Gaming 3

AMD Ryzen 7 1700

G.SKILL NT Series 16GB

LSI 6Gbps SAS HBA LSI 9201-8i

IBM 46M0997 ServeRAID Expansion Adapter 16-Port SAS Expander

Norco RPC-4220 case/backplane

tower-diagnostics-20190423-1415.zip

Is your mover running during these lockups?

35 minutes ago, IamSpartacus said:

Is your mover running during these lockups?

no, i have mover currently only running once per month. been seeing these lockups every few days.

Edited April 23, 20197 yr by drawde

i thought something was wrong with my plex install since that was the docker that i kept having to restart. i wiped out all the databases and metadata and had it rescrape/rebuild everything. Doesn't seem to have solved the issue.

 

i've also disabled c-states in my bios since apparently that's a thing for gen 1 ryzen.

 

i'm starting to think that high cpu usage overall is causing docker to hang/become unresponsive at times. when my CPU utlization is high (nzbget downloading/unpacking something while plex is transcoding), all my dockers slow down almost to a halt. oddly i didn't have this issue on my phenom II (passmark 3.5k) compared to the ryzen which is 13.7k.

 

to add some additional information, although not sure if important, i monitor my system with observium and the SNMP plugin, just to output a graph to a small status page i have. During these "lockups" there are gaps in data, i'm guessing because either the observium docker or SNMP plugin (or both) stops working along with most of my other dockers. when CPU goes down it will typically start again. no errors or anything in the syslog during these times.

 

example: 

 

any thoughts? could it be the I/O that is causing issues? should i move my appdata to an SSD?

Edited May 2, 20197 yr by drawde