Plex Media Server power usage constantly very high with no activity

Did you think of the process dumps or are these too large?

Here is a link to the memory dump.
https://drive.google.com/file/d/1VLzdiPxe8VHu4UJS9JPsrNj4V74bVu-Q/view?usp=sharing

To which set of log files does it belong?

These are the dump after I downgraded the server to version 1.13 (I forgot to dump before downgrading). If needed I could upgrade, dump, then downgrade again.

Since the issue appears with 1.15, we need unfortunately the dumps from that version.
Please do also fetch fresh log files when you grab the dump.

Here is memory dump for version 1.15 with excessive cpu usage. Logs coming next.

Here is link to dump:

https://drive.google.com/file/d/1kFLcSbv9fHCovfZSFmse8aUpz7tNYJgk/view?usp=sharing

Here are the logs for version 1.15.Plex Media Server Logs_2019-03-02_15-13-42.zip (4.7 MB)
Plex Media Server Logs_2019-03-02_15-14-13.zip (4.8 MB)

Why two files?
Again with and without the Security appliance? If so, did the issue arise even without the appliance?

First log was verbose, second not.

Thanks for the logs and the 1.15.1.710 dump.

You compared this with 1.13.9 - does this mean the same issue also exists on 1.14.1.5488?

The dump suggests that it is to do with SSDP discovery and handling multicasts and udp responses from the network devices. A thread clocked up 129 seconds out of the total 145 seconds Plex Media Server was running with - an observation: I feel it is premature to dump the process just after 145 seconds of launching Plex Media Server when we are looking for evidence of sustained high cpu when the server is idle. I feel that 145 seconds may have been too soon.

Also the tests on 1.15.1.710 appear to be giving different results with regards to SSDP Discovery in your different tests. Was the high cpu in all the 1.15.1.710 tests ? Was there a difference when BitDefender was not present?

Have a look at the discovered devices

1.13.9 Period 13:50:40 to 13:52:01

192.168.1.1
192.168.1.108
192.168.1.111
192.168.1.118
192.168.1.122
192.168.1.125
192.168.1.138
192.168.1.158
192.168.1.162
192.168.1.172
192.168.1.176
192.168.1.178
192.168.1.184
192.168.1.200
192.168.1.22
192.168.49.1

1.15.1.710 - No Bitdefender - Period 12:47:49 to 13:12:54

192.168.1.1
192.168.1.108
192.168.1.111
192.168.1.118
192.168.1.138
192.168.1.158
192.168.1.176
192.168.1.178
192.168.1.184
192.168.1.200
192.168.1.22
192.168.1.3
192.168.1.46
192.168.1.76
192.168.49.1

1.15.1.710 + BitDefender - Period 13:17:29 to 13:17:41

192.168.1.1
192.168.1.111
192.168.1.118
192.168.1.138
192.168.1.22
192.168.1.76

and last test with dump

1.15.1.710 SSDP period 15:03:42 to 15:03:54 (Dump at 15:06)

192.168.1.1
192.168.1.108
192.168.1.111
192.168.1.125
192.168.1.138
192.168.1.176
192.168.1.22

I would like to see more dumps to see if the cppnetlib stack gives more info - also would like to see performance chart (through task manager) showing the cpu and for it to be after the server has been started for say 10 minutes and period when the actions that happen on launch would have all finished and we enter a true quiescent state with no streaming, no scanning and no DLNA (you have already disabled this)

If the issue is to do with how a specific response from a device is handled, then having wireshark capture would help as well as the dump. If you manage to get that as well - you can send the pcap file to me and @OttoKerner by private message - would also need corresponding logs and dmp.

Have a question on the last set of diagnostics

The server was started on 1.15.1.710 at 15:03:38
At 15:06:03 the process dump was taken

The log shows a gap between 15:06 and 15:08 . Did it take 2 minutes to create the dump?

Every version of PMS after 1.13.9 has constant excessive cpu usage (9% - 10% minimum for weeks). With Bitdefender Box 2 removed from my system all version did not have the excessive cpu usage. Also Bitdefender Box 2 is the DNS server when installed and when I remove it I had to replace it with a router to act as DNS server so address would not necessarily match up. What exactly do you need me to do to help with the trouble shooting? Please list configurations, times and logs you want to see.

Thanks

Thanks but Version 1.15.2.793 did not fix the problem, still have high activity when nothing is playing.

Looked at the dmp I suspect a device on the network is either sending too many responses to SSDP search or some xml that is leading to a loop when being processed

Wireshark capture together with dmp and logs would hopefully shed some light on this

May be it is Bitdefender box that is sending these packets?
You say that without it, there is no issue. So need to establish what it is doing that is affecting Plex Media Server SSDP searches and processing of responses

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.