Andrew Wong created KUDU-2925:
---------------------------------
Summary: ToolTest.TestLoadgenZeroRowsPerThread didn't write to the
WAL in time
Key: KUDU-2925
URL: https://issues.apache.org/jira/browse/KUDU-2925
Project: Kudu
Issue Type: Improvement
Components: consensus, log, test
Affects Versions: 1.11.0
Reporter: Andrew Wong
Attachments: kudu-tool-test.3.txt.gz
I saw a run of ToolTest.TestLoadgenZeroRowsPerThread that failed because
writing to the WAL took 90s. It's not clear why, or whether this was due to
hardware issues or a deadlock somewhere.
{code:java}
I0819 03:23:52.093103 7689 catalog_manager.cc:3471] Sending
DeleteTablet(TABLET_DATA_DELETED) for tablet 1591b2416d5640d4a1f089ed959128c8
on a3a91c5fde7c49bcb22a8c908d76d428 (127.7.112.193:32983) (Table deleted at
2019-08-19 03:21:51 UTC)
W0819 03:23:52.271399 7684 debug-util.cc:397] Leaking SignalData structure
0x1fdb4e0 after lost signal to thread 7985
W0819 03:23:52.271497 7684 kernel_stack_watchdog.cc:198] Thread 7985 stuck at
/data/somelongdirectorytoavoidrpathissues/src/kudu/src/kudu/consensus/log.cc:671
for 89430ms:
Kernel stack:
(could not read kernel stack)
User stack:
<Timed out: thread did not respond: maybe it is blocking signals>
W0819 03:23:52.378173 7985 log.cc:669] Time spent T
00000000000000000000000000000000 P ce4f88cf8a264e82852c46587c4dce11: Append to
log took a long time: real 90.538s user 0.000s sys 0.000s
I0819 03:23:52.389586 7619 external_mini_cluster.cc:1136] Killing
/tmp/dist-test-taskfg0lzN/build/debug/bin/kudu-master with pid 7680
I0819 03:23:52.394946 7619 test_util.cc:135]
-----------------------------------------------
I0819 03:23:52.395031 7619 test_util.cc:136] Had fatal failures, leaving test
files at
/tmp/dist-test-taskfg0lzN/test-tmp/kudu-tool-test.3.ToolTest.TestLoadgenZeroRowsPerThread.1566184909971373-7619
[ FAILED ] ToolTest.TestLoadgenZeroRowsPerThread (121696 ms)
{code}
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
