Andrew Wong has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/14642 )

Change subject: [master] KUDU-2904 Crash master on disk error
......................................................................


Patch Set 1:

(2 comments)

Test looks good overall! I'd try looping the new test in release mode to see 
about potential flakiness.

http://gerrit.cloudera.org:8080/#/c/14642/1/src/kudu/integration-tests/disk_failure-itest.cc
File src/kudu/integration-tests/disk_failure-itest.cc:

http://gerrit.cloudera.org:8080/#/c/14642/1/src/kudu/integration-tests/disk_failure-itest.cc@328
PS1, Line 328:   constexpr int kNumReplicas = 3;
nit: this can actually be 1, right? We don't really care about actually 
creating tablet replicas, so the less state we create the better.


http://gerrit.cloudera.org:8080/#/c/14642/1/src/kudu/integration-tests/disk_failure-itest.cc@342
PS1, Line 342:  for (int table_suffix = 0; table_suffix < 10; table_suffix++) {
             :     string table_name = Substitute("test-$0", table_suffix);
             :     LOG(INFO) << "Creating table " << table_name;
             :     std::unique_ptr<client::KuduTableCreator> 
table_creator(client_->NewTableCreator());
             :     auto client_schema = 
client::KuduSchema::FromSchema(GetSimpleTestSchema());
             :     ASSERT_OK(table_creator->table_name(table_name)
             :                   .schema(&client_schema)
             :                   .set_range_partition_columns({ "key" })
             :                   .num_replicas(kNumReplicas)
             :                   .wait(true)
             :                   .Create());
Should we inject some sleeps in here so we're guaranteed to flush between table 
creations? I can imagine if the tables are created quickly and we might wind up 
with a single MRS, and then no compactions are possible.



--
To view, visit http://gerrit.cloudera.org:8080/14642
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I693eb7092c0b5feb530fb011937e636b40534495
Gerrit-Change-Number: 14642
Gerrit-PatchSet: 1
Gerrit-Owner: Bankim Bhavsar <[email protected]>
Gerrit-Reviewer: Adar Dembo <[email protected]>
Gerrit-Reviewer: Andrew Wong <[email protected]>
Gerrit-Reviewer: Bankim Bhavsar <[email protected]>
Gerrit-Reviewer: Kudu Jenkins (120)
Gerrit-Comment-Date: Tue, 05 Nov 2019 22:59:51 +0000
Gerrit-HasComments: Yes

Reply via email to