Massive RAM Usage in Activity Dashboard

Server Version#: 1.14.1.5488
Player Version#: Web 3.81.1

When viewing the plex dashboard, it takes about 30 seconds before it can display the bandwidth section. During this time RAM usage goes from a few hundred MB (average for websites) to sustaining 2.5GB, and reaching as high as 5.5GB.

Why is this happening?

1 Like

bump.

page is basically unusable. fix it please.

We’ll probably need more information, as it’s not trivially reproducible. I’ve had it here running for a few minutes and it’s not taking much memory at all.

I’m on Mojave 10.4.2 with release Safari.

Anything different over there? Any plugins/extensions?

I have no extensions enabled in Safari, and the only plugin on PleX is the Crunchyroll.bundle.

The plexmediaserver is running on a QNAP.

Here are my logs: https://mega.nz/#!jYkCFIAS!g_DRK2VLQOPtuOTjbSpTMAV0wdHlLdDKrIQsPxvAcgU

I don’t think the server logs are pertinent here (unless you’re seeing memory grow with only one server).

What version of OS and Safari?

macOS Mojave 10.14.2

Safari 12.0.2 (14606.3.4)

I’m only running a single server.

Trying to think about what else might be different between our two systems, and not coming up with any good ideas.

Anyone else with Safari seeing this issue?

@BrianGraham91 Could you open the browser developer tools and grab a screenshot of the network tab when this issue is happening?

1 Like

Here’s what happens when I load the page, the bandwidth request just sits there on refresh for a long time as the page is frozen

After that (luckily I had enough system memory) I let it sit until it failed or started again, and ended up with the following red lines:

Hmmm interesting. If you right click on the bandwidth request and open the URL in a new tab does it take a long time to load?

I have the same or similar issue with chrome+win10, memory and cpu usage goes up until the plex status page can be closed or navigated away from.

Google Chrome is up to date
Version 71.0.3578.98 (Official Build) (64-bit)

and plexweb

Version 1.14.1.5488 Check for Updates Up to date

Version 3.81.1

There is a large network activity burst going on for several seconds while the plexweb status page is loading, its visible with windows10 task manager performance tab.

Yes it does Moussa.

I also have this issue, on both my Win10 machine with Chrome at home as well as a Mac on Chrome at work. Both see massive CPU spikes when navigating to the dashboard that cripple the browser until the page is closed.

Browser at home:
Chrome Version 71.0.3578.98 (Official Build) (64-bit)

Interesting data point. Can you guys let us know what your servers are running (e.g. Intel Xeon, ARM) and what sort of network link separates browser from server? The theory is that it might be a combination of slower server and over-agrestic client.

I’m running a QNAP TS-563 w/ AMD GX-420MC SOC CPU w/ 16GB RAM. The QNAP is connected via ethernet to the router. Also the only plugin on my Plex is the Crunchyroll bundle.

My Mac Mini is connected via 5GHz WiFi to the router.

I host mine off of my old gaming rigs hardware after an upgrade. Plex is running as a Docker container in UNRAID. Generally performs pretty well. I see the issues on my wired LAN local to the server and at at work. I was doing some snooping around in the chrome dev tools but theyre not my specific area of IT expertise. I did notice that my previous statement that the whole browser is unresponsive seems to be inaccurate, it’s just the Plex tab that gets worked over by the CPU use. It seems like alot of the wait time on the page might be related to the following resources:

https://assets.plex.tv/deploys/desktop/env-eb2798cc3c7d9533df5b563963d5c394/3.81.1-e7d4f82/js/main-1-701bdda35867bcf9ad48-plex-3.81.1-e7d4f82.js

It also seems to reference ā€˜XHR Load’ taking lots of function call time in relation to that. Not sure if that helps.

System stats:
M/B: ASRock - P67 Extreme4
CPU: IntelĀ® Coreā„¢ i5-2500K CPU @ 3.30GHz
HVM: Enabled
IOMMU: [Disabled]
Cache: 1024 kB, 128 kB, 6144 kB
Memory: 12 GB (max. installable capacity 32 GB)

Thanks! And do you guys see the server process taking up lots of CPU during the process as well?

Yes, looking at the specific Chrome process in ProcExplorer for the affected tab shows the thread ā€œchrome.exe!GetHandleVerifier+0xaac10ā€ as the one hogging all of the CPU and when looking into its stack I get these results:

chrome_child.dll!ovly_debug_event+0xfca1f0
chrome_child.dll!ChromeMain+0x7a8d04
chrome_child.dll!ChromeMain+0x7a74cb
chrome_child.dll!ChromeMain+0x7a5744
chrome_child.dll!ChromeMain+0xa6e85
chrome_child.dll!ChromeMain+0xa6d63
chrome_child.dll!ChromeMain+0x85e1c7
chrome_child.dll!ChromeMain+0x85e0c3
chrome_child.dll!ovly_debug_event+0x1334682
chrome_child.dll!ovly_debug_event+0x12cd8bc

Again, not sure if useful, but datapoints I guess?

Sorry, I meant the server process, not Mr Chrome Tab. Apologize for not being more clear.

Apologies, it’s late and I can’t read apparently.

Looks like its a little spiky, but I’m not sure what is causing that. Doesn’t seem to be pegged out or anything.

Edit: Scratch that, it seems like when I closed the tab, those CPU spikes stopped.