CI Failure (serde overflow or underflow) in CompactionE2EIdempotencyTest.test_basic_compaction

[rystsov]

https://buildkite.com/redpanda/redpanda/builds/22032#0185fb59-d754-4805-9a3c-eca75b9bda6b

results/2023-01-29--001/CompactionE2EIdempotencyTest/test_basic_compaction/initial_cleanup_policy=compact.workload=Workload.TX/51

Module: rptest.tests.compaction_e2e_test
Class:  CompactionE2EIdempotencyTest
Method: test_basic_compaction
Arguments:
{
  "initial_cleanup_policy": "compact",
  "workload": "TX"
}

test_id:    rptest.tests.compaction_e2e_test.CompactionE2EIdempotencyTest.test_basic_compaction.initial_cleanup_policy=compact.workload=Workload.TX
status:     FAIL
run time:   1 minute 9.720 seconds

    <BadLogLines nodes=docker-rp-17(2) example="ERROR 2023-01-29 03:17:06,952 [shard 1] serde - serde.h:232 - Overflow or underflow detected when casting to nanoseconds, clamping to a limit. Input: -9223372036854775808  type: std::chrono::duration<long long, std::ratio<1, 1000>>">
Traceback (most recent call last):
  File "/usr/local/lib/python3.10/dist-packages/ducktape/tests/runner_client.py", line 135, in run
    data = self.run_test()
  File "/usr/local/lib/python3.10/dist-packages/ducktape/tests/runner_client.py", line 227, in run_test
    return self.test_context.function(self.test)
  File "/usr/local/lib/python3.10/dist-packages/ducktape/mark/_mark.py", line 476, in wrapper
    return functools.partial(f, *args, **kwargs)(*w_args, **w_kwargs)
  File "/root/tests/rptest/services/cluster.py", line 67, in wrapped
    self.redpanda.raise_on_bad_logs(allow_list=log_allow_list)
  File "/root/tests/rptest/services/redpanda.py", line 1740, in raise_on_bad_logs
    raise BadLogLines(bad_lines)
rptest.services.utils.BadLogLines: <BadLogLines nodes=docker-rp-17(2) example="ERROR 2023-01-29 03:17:06,952 [shard 1] serde - serde.h:232 - Overflow or underflow detected when casting to nanoseconds, clamping to a limit. Input: -9223372036854775808  type: std::chrono::duration<long long, std::ratio<1, 1000>>">

[ztlpn]

Right before the bad log line the following logs are produced:

TRACE 2023-01-29 03:17:06,952 [shard 1] tx - tm_stm_cache.cc:39 - looking for tx:{da9dc38a-1cf8-42fb-9477-068014ab6576} etag:1
TRACE 2023-01-29 03:17:06,952 [shard 1] tx - tm_stm_cache.cc:50 - looking for tx:{da9dc38a-1cf8-42fb-9477-068014ab6576} etag:1: can't find term:1 in _state
TRACE 2023-01-29 03:17:06,952 [shard 0] tx - tm_stm_cache.cc:39 - looking for tx:{da9dc38a-1cf8-42fb-9477-068014ab6576} etag:1
TRACE 2023-01-29 03:17:06,952 [shard 0] tx - tm_stm_cache.cc:50 - looking for tx:{da9dc38a-1cf8-42fb-9477-068014ab6576} etag:1: can't find term:1 in _state

This means that the messages are produced here - we tried searching for the transaction in tm_stm_cache, could not find it and return fetch_tx_reply(tx_errc::tx_not_found). Most probably the problem is that the fetch_tx_reply constructor from tx_ercc does not explicitly initialize the timeout_ms field and it gets filled with junk. This is a benign problem because this field in the errored fetch_tx_reply should not be used anywhere but still technically an UB so sev-medium.

[ztlpn]

Since there is no CompactionE2EIdempotencyTest in dev yet, I think it makes sense to fix this bug and test the fix as part of #8413

[rystsov]

@ztlpn the error is irrelevant to the PR and it shouldn't block it, I'll add ok_to_fail

[mmaslankaprv]

We need to default initialize the duration fields

[dotnwat]

We need to default initialize the duration fields

which fields?