-
Notifications
You must be signed in to change notification settings - Fork 553
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
rptest: use 10 nodes in OMB validation tests #16028
rptest: use 10 nodes in OMB validation tests #16028
Conversation
Drop the assumed number of nodes from 12 to 10 in OMB validation tests as that's the default number of nodes in duck.py and also the number we want to standardize on for all HTT tests.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Assuming there is still enough swarm workers to reach the desired connection count for T5+ then this LGTM.
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018cefe3-f837-465d-b05b-e63c4f633639:
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018cefe3-f82d-4d82-b86b-a89ab72fbdcd:
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018cefe3-f831-45a0-863c-224246594d66:
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018ceff4-fb26-4ae8-a327-7a343371e19a:
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018ceff4-fb23-4aa9-a514-ab050ad6f0b3:
new failures in https://buildkite.com/redpanda/redpanda/builds/43609#018ceff4-fb20-4ab9-9792-3b713d725002:
|
ducktape was retried in https://buildkite.com/redpanda/redpanda/builds/43609#018cefe3-f831-45a0-863c-224246594d66 ducktape was retried in https://buildkite.com/redpanda/redpanda/builds/43722#018cfe41-ce4e-40f4-806f-3a91864bd13f |
@@ -320,14 +325,14 @@ def test_max_partitions(self): | |||
self.redpanda, | |||
"ACK_ALL_GROUP_LINGER_1MS_IDEM_MAX_IN_FLIGHT", | |||
(workload, validator), | |||
num_workers=10, | |||
num_workers=self.CLUSTER_NODES - 1, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Does OMB make use of uneven nodes in "ensemble" mode?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, it assigns any "odd" worker to the consumer side, which I guess is good for us since we use 3:1 and 2:1 ratios, so maybe consumers are more likely to be overloaded (not in the many-partitions case though I guess due to the issue with high CPU in the drain loop there).
int numberOfProducerWorkers = extraConsumerWorkers ? (workers.size() + 2) / 3 : workers.size() / 2;
List<List<String>> partitions = Lists.partition(Lists.reverse(workers), workers.size() - numberOfProducerWorkers);
this.producerWorkers = partitions.get(1);
this.consumerWorkers = partitions.get(0);
(AFAIK we don't set extraConsumerWorkers
)
There should be, I tested with 10 nodes when I was doing T5 but not necessarily with the exact same machine type as we are suggesting changing to. Overall it seems like since we are changing the machine type and count, and all the tests are currently disabled in CI, we will have to run (through BK I guess, not manually) at some point all the various tiers again but for now we can stuff in as many possibly-invalidating changes as possible so that we only have to do that once at the end. |
/ci-repeat |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
/ci-repeat |
/backport v23.3.x |
Drop the assumed number of nodes from 12 to 10 in OMB validation tests as that's the default number of nodes in duck.py and also the number we want to standardize on for all HTT tests.
Backports Required
Release Notes