Adar Dembo has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/11448 )

Change subject: [tests] test scenario for KUDU-2580
......................................................................


Patch Set 2:

(5 comments)

http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc
File src/kudu/integration-tests/authn_token_expire-itest.cc:

http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc@106
PS2, Line 106:                             int num_tablet_servers,
             :                             int num_masters)
Nit: if you're already changing the argument order, prefer num_masters before 
num_tservers; that's how it's usually done in other tests, so this different 
order is error prone.


http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc@431
PS2, Line 431:       
Substitute("--leader_failure_max_missed_heartbeat_periods=$0",
             :           master_leader_failure_max_missed_heartbeat_periods_),
             :       Substitute("--raft_heartbeat_interval_ms=$0",
             :           master_raft_hb_interval_ms_),
             :       Substitute("--rpc_default_keepalive_time_ms=$0",
             :           master_rpc_keepalive_time_ms_),
Can you justify these customized settings?


http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc@437
PS2, Line 437:       "-v=1",
Was this just for debugging? Remove it now?


http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc@497
PS2, Line 497:     // Keep the connection to leader master open, time to time 
making requests
             :     // that go to the leader master, but not to other masters in 
the cluster.
Unless you disable failure detection in the masters, how can you guarantee that 
a new leader master isn't elected for some random reason? Will that mess up the 
test?


http://gerrit.cloudera.org:8080/#/c/11448/2/src/kudu/integration-tests/authn_token_expire-itest.cc@519
PS2, Line 519:       SleepFor(MonoDelta::FromMilliseconds(3 * 
master_raft_hb_interval_ms_));
Given that the whole thing is wrapped in ASSERT_EVENTUALLY, why do we need an 
explicit sleep?



--
To view, visit http://gerrit.cloudera.org:8080/11448
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I58218deef24cca7c524bc61700cd400cdaabd050
Gerrit-Change-Number: 11448
Gerrit-PatchSet: 2
Gerrit-Owner: Alexey Serbin <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Reviewer: Will Berkeley <[email protected]>
Gerrit-Comment-Date: Tue, 18 Sep 2018 17:00:21 +0000
Gerrit-HasComments: Yes

Reply via email to