Cannot run symbolicator tests #51241
Comments
|
I believe this looks rather like relay did not start in time, or according to the error message, failed to connect to postgres. |
This comment was marked as outdated.
This comment was marked as outdated.
|
Similar to #51471 |
|
You need to set symbolicator.enabled: true in your ~/.sentry/config.yml and rerun sentry devservices up. Those instructions are in all the files in tests/symbolicator but it would be better to have the tests detect there is no symbolicator and guide you on what to do. What CI is doing (host networking), is not what should be done locally in this case. Starting services via devservices locally only works with |
armenzg
changed the title
|
@joshuarli I have also followed the steps mentioned in the source code. My original comment was out of date; I've updated it. I have tried setting a new M2 machine and I hit the same problems. My work on #53643 is now being heavily impacted by this. |
|
I was able to reproduce (not the host.docker.internal stuff though, I get a different 2nd error) on Mac OS 13.3.1 on ARM, by forcing colima to use QEMU. I think neither Are you on Mac Intel? Show me your
|
|
I've switched to colima now. I get this error instead: |
|
Basically any test that relies on |
|
Armen I can't reproduce that... are you running on latest master without any modifications? |
|
Hi @joshuarli with clean master running on Venture 13.4: |
|
I migrated to colima and it failed too. |
@ashwoods - How long did you sleep for? I've increased the number of attempts, such that it waits quite a long time now, and still nothing. Here is what I do get: Test output============================================= test session starts ============================================== tests/symbolicator/test_unreal_full.py::SymbolicatorUnrealIntegrationTest::test_unreal_apple_crash_with_attachments ERROR [100%] ==================================================== ERRORS ==================================================== During handling of the above exception, another exception occurred: During handling of the above exception, another exception occurred: During handling of the above exception, another exception occurred: The above exception was the direct cause of the following exception: [many more copies of the r2d2 error] E 2023-10-19T18:34:27.658557Z ERROR r2d2: failed to lookup address information: Name or service not known [table creation and migration running] Applying social_auth.0002_default_auto_field... OK Regardless, I think I know the cause of the problem, though not yet entirely how to fix it. TL;DR, we can't handle it when the devserver or tests are run with a Things I discovered in my testing/wandering around the code:
So, what now? Option 1: Make everything work with arbitrary
Option 2: Make it so that running tests with the default
Naively, I would guess that option 2 is easier (and it would certainly make for better ux on the developer's end), but at this point I will defer to - and hand this off to - my dev-infra colleagues, as I don't have enough context to go beyond all of the above. @joshuarli - can I pass this back to you, please? [UPDATE] I pushed a branch with the changes I made, in case it's a helpful starting place. I hardcoded |
|
I have verified that if I don't use |
|
@armenzg is this still causing you problems? |
|
Hi, @jernejstrasner I have a slightly different issue. Possibly due to Colima (CC @joshuarli ) I got a new MBP M2 laptop in January I added |
I had to refresh some symbolicator tests for #51040 and I couldn't because I could not run the tests locally.
SENTRY_SNAPSHOTS_WRITEBACK=1 pytest -s -v tests/symbolicator -k test_full_minidump_invalid_extrawould fail for me withrelay did not start in time[1].This is now impacting me heavily on #53643. I've tried setting up an M2 machine and I have the same problem.
I did set
symbolicator.enabled: truein~/.sentry/config.yml.Running these commands to reproduce:
[1]
The text was updated successfully, but these errors were encountered: