Integration test step goes green even if actual tests fail #2922

Closed
opened 2018-05-08 01:42:26 +00:00 by meejah · 2 comments
meejah commented 2018-05-08 01:42:26 +00:00
Owner

Is maybe the "tox" command not propagating its exit-code .. "or something"? See e.g. https://travis-ci.org/tahoe-lafs/tahoe-lafs/jobs/375959463

Is maybe the "tox" command not propagating its exit-code .. "or something"? See e.g. <https://travis-ci.org/tahoe-lafs/tahoe-lafs/jobs/375959463>
tahoe-lafs added the
unknown
normal
defect
1.12.1
labels 2018-05-08 01:42:26 +00:00
tahoe-lafs added this to the undecided milestone 2018-05-08 01:42:26 +00:00
meejah commented 2018-05-08 02:28:54 +00:00
Author
Owner

Locally at least (Debian stretch) tox -e integration does correctly return 1 (if e.g. the "Chutney" tests fail, in the case I was looking at).

Locally at least (Debian stretch) `tox -e integration` does correctly return `1` (if e.g. the "Chutney" tests fail, in the case I was looking at).
exarkun commented 2020-01-17 17:04:57 +00:00
Author
Owner

Hm. I think someone clicked the rebuild button on that travis build. Travis re-uses build numbers! So when you rebuild a failed job you lose all of the state from the failed run and it is replaced by the results of the new run. Fun stuff.

This sounds like a terrible bug sadly all of the information associated with the failure is gone.

We should re-open this if we observe the problem again and include all relevant information on the ticket instead of linking out to travis to avoid losing it again.

FWIW, fortunately or unfortunately, I've seen failed integration test runs on CI... so this doesn't always happen. :/

Hm. I think someone clicked the rebuild button on that travis build. Travis re-uses build numbers! So when you rebuild a failed job you lose all of the state from the failed run and it is replaced by the results of the new run. Fun stuff. This sounds like a terrible bug sadly all of the information associated with the failure is gone. We should re-open this if we observe the problem again and include all relevant information on the ticket instead of linking out to travis to avoid losing it again. FWIW, fortunately or unfortunately, I've seen failed integration test runs on CI... so this doesn't *always* happen. :/
tahoe-lafs added the
cannot reproduce
label 2020-01-17 17:04:57 +00:00
exarkun closed this issue 2020-01-17 17:04:57 +00:00
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: tahoe-lafs/trac-2024-07-25#2922
No description provided.