Description
The HmsSentryConfigurations/AlterTableRandomized.TestRandomSequence/1 scenario of the alter_table_randomized-itest.cc exhibits flakiness in case of TSAN builds, time to time failing with errors like below:
F0719 06:51:55.884040 245 alter_table-randomized-test.cc:499] Check failed: _s.ok() Bad status: Not found: The specified column does not exist
It's pretty clear what happened: the call timed out as it's seen from the client side,
W0719 06:51:55.871377 445 rpcz_store.cc:253] Call kudu.master.MasterService.AlterTable from 127.0.0.1:54308 (request call id 178) took 10022 ms (10 s). Client timeout 9999 ms (10 s)
and the client retried performing the same operation:
W0719 06:51:55.850235 874 master_proxy_rpc.cc:192] Re-attempting AlterTable request to leader Master (127.0.61.126:33771)
but altering the table (dropping column c486) has actually succeeded:
W0719 06:51:55.868930 493 hms_client.cc:272] Time spent alter HMS table: real 9.997s user 0.000s sys 0.001s
I0719 06:51:55.890976 990 tablet.cc:1259] T 7f145a02995242298c013c329420b6f5 P d12954c90bf3483f80f0222ee1a74ef9: Alter schema from ( 10:key INT32 NOT NULL, 11:c486 INT32 NULLABLE, PRIMARY KEY (key) ) version 1 to ( 10:key INT32 NOT NULL, 11:c746 INT32 NULLABLE, PRIMARY KEY (key) ) version 2
I'm attaching full test log reference.
Attachments
Attachments
Issue Links
- relates to
-
KUDU-3215 Flakiness in HmsConfigurations/AlterTableRandomized.TestRandomSequence
- Open