> On April 14, 2017, 12:57 a.m., Alexander Kolbasov wrote: > > sentry-provider/sentry-provider-db/src/main/java/org/apache/sentry/service/thrift/SentryService.java > > Line 196 (original), 186 (patched) > > <https://reviews.apache.org/r/58221/diff/9/?file=1692270#file1692270line198> > > > > Can you clarify why it should be started in constructor and why would > > tests fail otherwise? > > Na Li wrote: > the test TestDbSentryOnFailureHookLoading setup calls > AbstractTestWithDbProvider.createContext(), which expects the context to be > not null after creating SentryService and before calling > SentryService.start(). If we don't schedule the cleaner in constructor, we > will fail at assumeNotNull(context); if we schedul the cleaner in > constructor, it does not fail at assumeNotNull(context); > > AbstractTestWithDbProvider.createContext() > { > ... > assumeNotNull(context); > context = AbstractTestWithHiveServer.createContext(properties); > policyFile > .setUserGroupMapping(StaticUserGroup.getStaticMapping()) > .write(context.getPolicyFile(), policyFilePath); > > startSentryService(); > ... > } > > Alexander Kolbasov wrote: > That's rather weird - the context is created right after you verify that > it is not null. How can that work? And how does the cleaner service affects > this strange assert?
Now, it is not started at constructor. So the issue is removed. > On April 14, 2017, 12:57 a.m., Alexander Kolbasov wrote: > > sentry-provider/sentry-provider-db/src/main/java/org/apache/sentry/service/thrift/SentryService.java > > Lines 425 (patched) > > <https://reviews.apache.org/r/58221/diff/9/?file=1692270#file1692270line454> > > > > The code uses synchronized() on both, so the comment isn't quite right > > - it would handle concurrent calls to start()/stop() - not that it is > > needed. > > Na Li wrote: > the synchronized is orignal. You mentioned if multiple threads accese it > at the same time, since we don't protect variables, it is still not thread > safe. > > Na Li wrote: > do you want me to remove the comment? > > Alexander Kolbasov wrote: > It may be better to update it - the current comment contradicts the code. I have removed the comment that it is not thread safe - Na ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/58221/#review171970 ----------------------------------------------------------- On April 18, 2017, 3:46 a.m., Na Li wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/58221/ > ----------------------------------------------------------- > > (Updated April 18, 2017, 3:46 a.m.) > > > Review request for sentry, Alexander Kolbasov, Hao Hao, kalyan kumar > kalvagadda, and Sergio Pena. > > > Repository: sentry > > > Description > ------- > > SENTRY-1649 move HMS follower to runServer > > > Diffs > ----- > > > sentry-provider/sentry-provider-db/src/main/java/org/apache/sentry/service/thrift/HMSFollower.java > 16676fb > > sentry-provider/sentry-provider-db/src/main/java/org/apache/sentry/service/thrift/SentryService.java > 132db63 > > > Diff: https://reviews.apache.org/r/58221/diff/15/ > > > Testing > ------- > > > Thanks, > > Na Li > >
