rptest/scale_tests: compute expected time for non-ts workload #16704
Conversation
nvartolomei
force-pushed
the
nv/scale-mpt-fix-non-tiered-storage
branch
from
31cb4d0
to
169f6af
Compare
|
/cdt |
nvartolomei
force-pushed
the
nv/scale-mpt-fix-non-tiered-storage
branch
from
169f6af
to
d0fe469
Compare
|
/cdt |
|
Test failures are unrelated. Some CI infra instability and builds/ducktape runs are irrelevant for this runs beside cdt-aws-ci. |
| @@ -587,12 +587,22 @@ def _write_and_random_read(self, scale: ScaleParameters, topic_names): | |||
| self.logger.info( | |||
| "Write+randread stress test complete, verifying sequentially") | |||
|
|
|||
| max_msgs = None | |||
| expect_transmit_time = 600 # empirically derived | |||
There was a problem hiding this comment.
Ah my comment got swallowed, though this still LGTM.
I would've expected this to be set to None if trying to revert the behavior to before https://github.com/redpanda-data/redpanda/pull/15287/files#diff-1918605fe4f7ee419c1af283695cfb29917fbad61e10faafa40623c051a5e0c4L590
That said, looking around, I'm not sure where None is handled gracefully, and 600 seems much better
There was a problem hiding this comment.
You are right. I'd leave it as-is as explicit value here is a tad more readable.
|
Thanks @nvartolomei ! |
This restores the default timeout for non-ts workload as it was pre my change. We have the bandwidth setting for the single producer/consumer but we fail to achieve it with random reads.
Fixes #16696
Backports Required
Release Notes