High CPU synoscgi_SYNO.Core.Package_2_list

[jucor]

Hi everyone!

For the last few hours, my Syno DS218+ (DSM 6.2.3-25426 Update 3) CPU is going through the roof, maxing at 100% for several minutes, going down to 20%, and back up to 100%, although idle. It's killing me. No Download station, no docker, no VM. No Plex transcoding either. Noone logged on Plex, actually.


htop and Resource Monitor do not show a single very large process gone wild, except sometimes indeed `synoscgi_SYNO.Core.Package_2_list` going to 38%. No other massive process.

I'm stumped. I have tried rebooting, no change. I don't think I have updated anything today either.
Any idea, please?
Thanks !

Here's realtime CPU over the last few minutes:

and over 1 day (2-4am is the nightly borgbackup job, that's normal) with the huge spikes this evening:

and finally over 1 week for context (with the nightly borgbackup jobs, very periodic pattern, and today's mess):

 

[Robbie]

Suggest you SSH into the NAS and use the 'top' command to see what is going on. Sometimes disk issues can manifest themselves as unwarranted CPU activity so keep an open mind.

 

[jucor]

Thanks Robbie. Yes, I didn't screenshot it but have been using htop (a slightly more advanced version of top) over SSH, and that's how I obtained the name of the process in the title.

I tried to use iotop from opkg but it crashed on me. Any good other CLI tool to monitor I/o?

 

[Rusty]

Just to be clear, there is no data scrubbing going on? Storage Manager status of storage pool/volume?

 

[jucor]

I just checked:
- no data scrubbing
- status of pool and volumes: Healthy.

Weirdly, *lots* of log entries "Link up" "Link Down", coinciding with the CPU usage peaks. Do not know which is cause, which is consequence See aligned screenshots below:

 

[Rusty]

I just checked:
- no data scrubbing
- status of pool and volumes: Healthy.

Weirdly, *lots* of log entries "Link up" "Link Down", coinciding with the CPU usage peaks. Do not know which is cause, which is consequence See aligned screenshots below:

View attachment 3146

The fact that you have a "user" utilization, would suggest it is def an app/service that is on top of the system itself.

Can't say what exactly, but something is really kicking it. At 50% user utilization would suggest a single core out of those 2 that 218+ has is being used to the max. Could be a single-threaded process of some sort. Wouldn't put my money on the network adapter is up/down.

 

[jucor]

Thanks Rusty. This is weird, the log entries on the link up/down started about the same time as the CPU usage going bersek, around 7pm yesterday (see screenshot below). But clearly that might be a consequence of a user app kicking it.

There's no clear culprit from htop (and things change all the time), so I'll try with a performance monitor that aggregates over time, and report.

-- post merged: 25. Feb 2021 --

Found the culprit with `nmon`! It's Plex. Will now check *those* logs, and see what's going on.
Do you recommend Plex via Docker over Plex via their "official" synology package?

 

[Rusty]

Do you recommend Plex via Docker over Plex via their "official" synology package?

It shouldn't matter. Both can work really well. I have it on bare metal (package center install) because I have had it running for years (before docker on syno) and it works just fine. Also, the Docker version did not always support HW transcoding so never migrated to a docker version myself.

Today it is different and I see no reason not to have it in a docker setup.

 

[jucor]

I am stumped. I deactivated pretty much everything, but still have long spikes of CPU usage. I recorded with atop, and I see moments were the CPU is blowing off, but no process seems to be using it to that extend -- see screenshot.
Any idea why the CPU usage could be sky high but no process shown exploding, please?