Tim Armstrong has posted comments on this change. ( http://gerrit.cloudera.org:8080/8490 )
Change subject: IMPALA-2248: Make idle_session_timeout a query option ...................................................................... Patch Set 10: (8 comments) http://gerrit.cloudera.org:8080/#/c/8490/10//COMMIT_MSG Commit Message: http://gerrit.cloudera.org:8080/#/c/8490/10//COMMIT_MSG@7 PS10, Line 7: IMPALA-2248: Make idle_session_timeout a query option > * Yeah, currently it doesn't have any effect for the shell. As you said, it Sounds reasonable. We'll have to make sure to document the limitations clearly - we can file a Doc JIRA as a follow-up. http://gerrit.cloudera.org:8080/#/c/8490/10/be/src/service/client-request-state.cc File be/src/service/client-request-state.cc: http://gerrit.cloudera.org:8080/#/c/8490/10/be/src/service/client-request-state.cc@207 PS10, Line 207: UpdateSessionTimeout(atoi(value.c_str())); > At this point the value can be invalid, e.g. the client wants to set the ti Right, but if the client provides an invalid value, I think it will just get swallowed up here. E.g. I tried set idle_session_timeout="foo"; and it seems to fail silently. I'd expect it return the "Invalid idle session timeout: " error to the client. I think we should add tests to set.test for validation of the option. http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java File fe/src/test/java/org/apache/impala/service/JdbcTest.java: http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java@616 PS11, Line 616: ew Long( nit: is the new Long() necessary? Seems like it should be automatically cast http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java@626 PS11, Line 626: int sleepPeriod = (int)(timeout * timeoutTolerance * 1000) + 500; I think the wakeup timing could do with a brief explanation. Is the idea is that each iteration might let one session expire and then renews the timeout for the remaining sessions? http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java@634 PS11, Line 634: new Long same here http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java@636 PS11, Line 636: Boolean bool? Not sure why we need to use the boxed type http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/service/JdbcTest.java@639 PS11, Line 639: try(ResultSet rs = connection.createStatement().executeQuery("SELECT 1+2")) { nit: whitespace http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/util/Metrics.java File fe/src/test/java/org/apache/impala/util/Metrics.java: http://gerrit.cloudera.org:8080/#/c/8490/11/fe/src/test/java/org/apache/impala/util/Metrics.java@43 PS11, Line 43: public Object getMetric(String metric) throws Exception { A brief comment would be helpful, mainly to explain what it returns - when reading the test I was initially confused about the casting to (Long). -- To view, visit http://gerrit.cloudera.org:8080/8490 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-MessageType: comment Gerrit-Change-Id: I32e2775f80da387b0df4195fe2c5435b3f8e585e Gerrit-Change-Number: 8490 Gerrit-PatchSet: 10 Gerrit-Owner: Zoltan Borok-Nagy <borokna...@cloudera.com> Gerrit-Reviewer: Attila Jeges <atti...@cloudera.com> Gerrit-Reviewer: Csaba Ringhofer <csringho...@cloudera.com> Gerrit-Reviewer: Dan Hecht <dhe...@cloudera.com> Gerrit-Reviewer: Gabor Kaszab <gaborkas...@cloudera.com> Gerrit-Reviewer: Laszlo Gaal <laszlo.g...@cloudera.com> Gerrit-Reviewer: Michael Ho <k...@cloudera.com> Gerrit-Reviewer: Thomas Tauber-Marshall <tmarsh...@cloudera.com> Gerrit-Reviewer: Tim Armstrong <tarmstr...@cloudera.com> Gerrit-Reviewer: Zoltan Borok-Nagy <borokna...@cloudera.com> Gerrit-Comment-Date: Tue, 05 Dec 2017 01:25:00 +0000 Gerrit-HasComments: Yes