rpminspect tests may require a lot of memory? #545

New issue

Open

opened 2026-02-17 14:22:49 +00:00 by lecris · 3 comments

lecris commented

2026-02-17 14:22:49 +00:00

Owner

Hit this issue on rpminspect intel-igc

[2026-02-16 20:06:50] [W] [Thread-1] [oom] Reservation memory 4096 MiB reached
[2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] trying to acquire pipeline cancellation lock
[2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] acquired pipeline cancellation lock
[2026-02-16 20:08:30] [E] [Thread-1] [oom] Terminating pipeline because worker is out of memory, more than 8192 MiB consumed.
[2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] released pipeline cancellation lock
[2026-02-16 20:08:30] [W] Signal SIGUSR2 received
[2026-02-16 20:08:30] [W] Sending SIGKILL to child process 'tmt' (PID 184)
[2026-02-16 20:08:30] [W] [worker_0] [stderr] [None:x86_64:/rpminspect]             duration limit: 43200 seconds
[2026-02-16 20:08:30] [+] ---^---^---^---^---^--- End of command output
[2026-02-16 20:08:30] [+] [worker_0] [test-schedule-tmt] tmt exited with code -9

I see it is a consistent issue with that package

Hit this issue on rpminspect `intel-igc` ``` [2026-02-16 20:06:50] [W] [Thread-1] [oom] Reservation memory 4096 MiB reached [2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] trying to acquire pipeline cancellation lock [2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] acquired pipeline cancellation lock [2026-02-16 20:08:30] [E] [Thread-1] [oom] Terminating pipeline because worker is out of memory, more than 8192 MiB consumed. [2026-02-16 20:08:30] [+] [Thread-1] [testing-farm-request] released pipeline cancellation lock [2026-02-16 20:08:30] [W] Signal SIGUSR2 received [2026-02-16 20:08:30] [W] Sending SIGKILL to child process 'tmt' (PID 184) [2026-02-16 20:08:30] [W] [worker_0] [stderr] [None:x86_64:/rpminspect] duration limit: 43200 seconds [2026-02-16 20:08:30] [+] ---^---^---^---^---^--- End of command output [2026-02-16 20:08:30] [+] [worker_0] [test-schedule-tmt] tmt exited with code -9 ``` I see it is a consistent issue with that package

lecris added the

bug

area/rpminspect

area/Testing Farm

labels

2026-02-17 14:22:49 +00:00

mvadkert commented

2026-02-18 01:25:50 +00:00

Owner

@lecris yeah, this is a known issue for certain components, although I am a bit suprised intel-igc hits it. Could be also a rpminspect (or some of the tools it uses) bug. Should really happen only for components with large code bases.

Anyway, I was thinking how to solve on our side, but I am getting to a conclusion for some components we should just run this on a VM instead of container. I can try to draft it in next sprint. It would need to create another rpminspect plan that would run against a beefy VM, which would run the rpminspect container, instead of running the rpminspect container directly on Testing Farm workers (which is superb for most of the cases as it is lightweight and thus faster).

@lecris yeah, this is a known issue for certain components, although I am a bit suprised `intel-igc` hits it. Could be also a rpminspect (or some of the tools it uses) bug. Should really happen only for components with large code bases. Anyway, I was thinking how to solve on our side, but I am getting to a conclusion for some components we should just run this on a VM instead of container. I can try to draft it in next sprint. It would need to create another rpminspect plan that would run against a beefy VM, which would run the rpminspect container, instead of running the rpminspect container directly on Testing Farm workers (which is superb for most of the cases as it is lightweight and thus faster).

lecris commented

2026-02-18 15:09:34 +00:00

Author

Owner

Was this discussed with rpminspect if this is not something that they could fix? It does not actually show which test it failed at. Did it fail somewhere else, maybe at the rpm download?

Otherwise, reworking it to use VM, would make the rpminspect_runner image obsolete (not opposed to that).

Was this discussed with rpminspect if this is not something that they could fix? It does not actually show which test it failed at. Did it fail somewhere else, maybe at the rpm download? Otherwise, reworking it to use VM, would make the `rpminspect_runner` image obsolete (not opposed to that).

lecris added this to the Board project

2026-06-10 12:35:50 +00:00

lecris added the

Priority

Medium

label

2026-06-10 12:35:59 +00:00

lecris commented

2026-06-10 12:36:44 +00:00

Author

Owner

Will address in a V2 rewrite in https://github.com/packit/tmt-plans/pull/37