Either use script from PR or run test_topic_swarm test while selecting number of topics near the limit of the cluster abilities: (<rp_nodes_cluster> * <vcpu_count> * 1000) / <replica_count>. Example for i3en.xlarge: (941000)/3 = 12000
Run the test and check BadLogLines errors at the end
Additional information
Backtrace decoding for one of the logs.
cmd: python3 ~/seastar-addr2line.py -v -a /usr/bin/llvm-addr2line-14 -e /opt/redpanda/libexec/redpanda -f ~/tests/results/2024-02-07--001/ManyPartitionsTest/test_topic_swarm/1/RedpandaService-0-140592323297680/ip-172-31-3-143/redpanda.log
Version & Environment
Manual CDT run using 10 i3en.xlarge nodes on AWS.EC2
Redpanda version: (use
rpk version
):Kafka python client
What went wrong?
When creating large topic number (11950) on 9 node redpanda cluster test failed with BadLogLines:
What should have happened instead?
Redpanda able to create topics up to the limit and/or report something like "No more topics creation possible, limit reached"
How to reproduce the issue?
Ducktape cmd line is
(<rp_nodes_cluster> * <vcpu_count> * 1000) / <replica_count>
. Example for i3en.xlarge: (941000)/3 = 12000Additional information
Backtrace decoding for one of the logs. cmd:
python3 ~/seastar-addr2line.py -v -a /usr/bin/llvm-addr2line-14 -e /opt/redpanda/libexec/redpanda -f ~/tests/results/2024-02-07--001/ManyPartitionsTest/test_topic_swarm/1/RedpandaService-0-140592323297680/ip-172-31-3-143/redpanda.log