adelapena commented on a change in pull request #870:
URL: https://github.com/apache/cassandra/pull/870#discussion_r574054070
##########
File path:
test/unit/org/apache/cassandra/cql3/validation/operations/AlterTest.java
##########
@@ -269,6 +276,63 @@ public void testCreateAlterKeyspaces() throws Throwable
"max_threshold", "32")));
}
+ @Test
+ public void testCreateAlterKeyspacesRFWarnings() throws Throwable
+ {
+ requireNetwork();
+ Thread.sleep(1000); // Hack: requireNetwork() will clear warnings
async so give some time to complete
Review comment:
> Thx Andres for multiplexing it but running the method individually
won't repro the issue. I get it, sometimes, only if running the whole class. It
failed only 1 out of 20 for me today whereas it used to fail every time some
days ago...
The jobs above also include [100
runs](https://jenkins-dse.build.dsinternal.org/view/Parameterized/job/parameterized-testall/691/)
for the whole class.
[Here](https://jenkins-dse.build.dsinternal.org/view/Parameterized/job/parameterized-testall/692/)
and
[here](https://jenkins-dse.build.dsinternal.org/view/Parameterized/job/parameterized-testall/693/)
are two more runs of the whole class with 200 repetitions each. That totals
500 runs of the whole class without a single failure, so I'd say that
definitely we can't reproduce it with the multiplexer.
Unfortunately testing this with CircleCI is not that easy, but I have seen a
few full unit tests runs that pass without the sleep:
*
https://app.circleci.com/pipelines/github/adelapena/cassandra/175/workflows/3432945a-05bc-40b4-93cf-607b1ce6566c/jobs/1291
*
https://app.circleci.com/pipelines/github/adelapena/cassandra/175/workflows/4a879181-921a-47bf-8cd3-a886fbb99fa7/jobs/1297
*
https://app.circleci.com/pipelines/github/adelapena/cassandra/175/workflows/a39779a1-e059-4d5c-9563-3de78d9bd864/jobs/1303
*
https://app.circleci.com/pipelines/github/adelapena/cassandra/175/workflows/544a5037-0261-49d0-b5c1-9dfdfac09bc5/jobs/1311
*
https://app.circleci.com/pipelines/github/adelapena/cassandra/175/workflows/1c58b779-7d80-4e79-b7d7-1b73aecc62b8/jobs/1314
I guess that the (in)famous poll, if needed, doesn't have to be for
`!warns.empty` but for whatever signals that `requireNetwork` has finished and
therefore it's safe to use. Said that, if we still think that the test can fail
without the wait and we don't want to spend more time investigating what's the
condition that signals the finalization of `requireNetwork` I think I can live
with the current sleep.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
