[mozproxy] Raptor/AWSY tests can leave mitmdump processes behind with each of those consuming 100% cpu load
Categories
(Testing :: Raptor, defect, P3)
Tracking
(firefox150 fixed)
| Tracking | Status | |
|---|---|---|
| firefox150 | --- | fixed |
People
(Reporter: whimboo, Assigned: kshampur)
References
(Blocks 1 open bug)
Details
(Whiteboard: [fxp])
Attachments
(2 files)
I noticed again a huge CPU load on my machine and when checking the activity monitor a couple of mitmproxy processes were running with each of them utilizing 100% of CPU load. See screenshot attached.
Most likely this happens when the Raptor job gets aborted eg. Ctrl+C, but maybe also in other situations.
This might have a really bad impact on our CI machines, especially the Mac Minis. We should get this investigated and fixed.
Updated•6 years ago
|
Updated•5 years ago
|
| Reporter | ||
Comment 2•5 years ago
|
||
I can also see this for AWSY now. Basically each time you break a running mach raptor/awsy test job with Ctrl+C a mitmdump instance is left behind. Usually it doesn't consume any CPU but can hold a quite amount of memory, e.g one of these instances locally uses 377MB! Then after a while or due to some other reason such processes go crazy and the reported CPU usage can be seen.
Comment 3•1 year ago
|
||
This is still a valid issue, most frequently reproducible on Android. Less frequently on Desktop platforms.
Updated•11 months ago
|
Updated•11 months ago
|
| Assignee | ||
Comment 4•10 months ago
|
||
| Assignee | ||
Comment 5•10 months ago
|
||
I added a WIP patch that seems to work for me. I think only edge case not accounted for is when Exceptions are hit, otherwise things like keyboard interrupt and session terminations are handled
| Assignee | ||
Updated•10 months ago
|
Updated•3 months ago
|
Updated•3 months ago
|
Updated•3 months ago
|
Updated•3 months ago
|
Updated•3 months ago
|
Description
•