You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am using this container for automated testing with Chrome controlled over CDP.
Occasionally for undetermined reasons and with no apparent pattern the container hangs and if resources (RAM, CPU) are not limited the entire host machine hangs: CPU goes to 100%, disk IO goes to 100%, RAM 100% and swap 100%.
I am forced to run the container without limiting its CPU because if I do performance degrades dramatically even if I assign 4 cores out of 8 for example. For some reason any kind of resource limiting degrades performance so I'm running it without limits.
I suspect it's some kind of memory leak as if I schedule hourly container restarts the hangs almost never happen (unless under heavier loads than usual and by that I mean 2-3 tabs open/concurrent tests running)
I am still investigating what's going on but it's difficult getting access to logs when the entire VM needs to be rebooted as the container is very verbose and have to set aside some time for log digging.
Expected Behavior
It should preferably crash rather than bogging down the entire VM.
Steps To Reproduce
Unknown. Just leave it running for hours/days under load.
Environment
GCP Container Optimized OS (comes with Docker preinstalled)
VMs with 4/6/8 cores, 8/16 Gb RAM
you need --shm-size=1gb at a minimum.
Please also test out the new Wayland mode, X11 side will not be getting new features. -e PIXELFLUX_WAYLAND=true .
I have never seen this personally but this is the second OOM report we have gotten.
I run chromium on a long run container, unfortunately I update it automatically and it is never up for more than a week.
I need a replication method to fix it, for all we know GCP does some kind of resource saving thing that causes this.
We've set it initially to 2 then 4 then 8Gb, doesn't seem to be a memory issue, besides we're not testing complex pages, there are a couple SPAs but their memory footprint is under 300Mb in prod with real data.
This will be a tough nut to crack but I'll keep playing with settings and keep looking for logs, hopefully will find something.
I suspect Wayland mode will just fix it, though it was designed with GPUs in mind the software mode is still on par with X11.
You can also clamp your X11 virtual display to like 4k. -e MAX_RESOLUTION=3840x2160
I suspect Wayland mode will just fix it, though it was designed with GPUs in mind the software mode is still on par with X11. You can also clamp your X11 virtual display to like 4k. -e MAX_RESOLUTION=3840x2160
Will try this tomorrow 🙏
Also a couple weeks ago we left a MBP on caffeine with the viewer open to watch the tests running and went out for coffee lol but when we got back the browser tab had eaten all the 32 gigs of RAM on the macbook and went like 20Gb more into swap, it had to be restarted. Didn't pay much attention to it but it may be related. Only happened once back then, couldn't figure out why.
You need to watch chrome's usage, when chrome is debugging or doing any network/performance profiling that portion of the codebase is not optimized.
Must be some kind of memory leak or something Chrome-related, right now after some 45 minutes of running tests it's idle but RAM is at 100% and swapping. It launched a lot of chrome processes for some reason, like 300 of them but there's only one window open with a single tab which is sitting on the default page. Nothing in the Chrome task manager either. I know the tests we run close the windows on completion, not sure why it would keep processes open
This issue has been automatically marked as stale because it has not had recent activity. This might be due to missing feedback from OP. It will be closed if no further activity occurs. Thank you for your contributions.
Activity
github-actions commentedon Feb 5, 2026
Thanks for opening your first issue here! Be sure to follow the relevant issue templates, or risk having this issue marked as invalid.
Roxedus commentedon Feb 5, 2026
This shows the signs of a unprotected instance that has been hijacked to cryptomine.
nicksterious commentedon Feb 5, 2026
We've considered that but negative, everything is firewalled, content is controlled.
thelamer commentedon Feb 5, 2026
you need
--shm-size=1gbat a minimum.Please also test out the new Wayland mode, X11 side will not be getting new features.
-e PIXELFLUX_WAYLAND=true.I have never seen this personally but this is the second OOM report we have gotten.
I run chromium on a long run container, unfortunately I update it automatically and it is never up for more than a week.
I need a replication method to fix it, for all we know GCP does some kind of resource saving thing that causes this.
nicksterious commentedon Feb 5, 2026
We've set it initially to 2 then 4 then 8Gb, doesn't seem to be a memory issue, besides we're not testing complex pages, there are a couple SPAs but their memory footprint is under 300Mb in prod with real data.
This will be a tough nut to crack but I'll keep playing with settings and keep looking for logs, hopefully will find something.
thelamer commentedon Feb 5, 2026
I suspect Wayland mode will just fix it, though it was designed with GPUs in mind the software mode is still on par with X11.
You can also clamp your X11 virtual display to like 4k.
-e MAX_RESOLUTION=3840x2160nicksterious commentedon Feb 5, 2026
Will try this tomorrow 🙏
Also a couple weeks ago we left a MBP on caffeine with the viewer open to watch the tests running and went out for coffee lol but when we got back the browser tab had eaten all the 32 gigs of RAM on the macbook and went like 20Gb more into swap, it had to be restarted. Didn't pay much attention to it but it may be related. Only happened once back then, couldn't figure out why.
thelamer commentedon Feb 5, 2026
You need to watch chrome's usage, when chrome is debugging or doing any network/performance profiling that portion of the codebase is not optimized.
nicksterious commentedon Feb 5, 2026
Must be some kind of memory leak or something Chrome-related, right now after some 45 minutes of running tests it's idle but RAM is at 100% and swapping. It launched a lot of chrome processes for some reason, like 300 of them but there's only one window open with a single tab which is sitting on the default page. Nothing in the Chrome task manager either. I know the tests we run close the windows on completion, not sure why it would keep processes open
Would a
chrome://restartafter each run help?thelamer commentedon Feb 5, 2026
If you are using it in test mode make sure the wrapped-chrome script is not messing with you cleaning up singletons.
LinuxServer-CI commentedon Mar 8, 2026
This issue has been automatically marked as stale because it has not had recent activity. This might be due to missing feedback from OP. It will be closed if no further activity occurs. Thank you for your contributions.
LinuxServer-CI commentedon Jun 6, 2026
This issue is locked due to inactivity