Re: [hibernate-dev] Missing transaction hangs the testsuite

From: "Steve Ebersole" <steve(a)hibernate.org&gt; To: "Sanne Grinovero" <sanne(a)hibernate.org&gt; Cc: "Hibernate.org" <hibernate-dev(a)lists.jboss.org&gt; Sent: Saturday, 18 July, 2015 12:04:01 AM Subject: Re: [hibernate-dev] Missing transaction hangs the testsuite So this is definitely not limited to 5.0. This happens in 4.3 as well. I pointed 4.3 locally to my MariaDB instance and it is hanging. In every case I have seen so far, this is problems in the test code though. On Fri, Jul 17, 2015 at 1:33 PM Steve Ebersole <steve(a)hibernate.org&gt; wrote: > Here is something else interesting... > > So at the point that the test hang against MySQL for me, I have: > > MariaDB [(none)]> SHOW FULL processlist; > > +-----+--------------------+-----------------+--------------------+---------+------+---------------------------------+--------------------------------+----------+ > | Id | User | Host | db | > Command | Time | State | Info > | Progress | > > +-----+--------------------+-----------------+--------------------+---------+------+---------------------------------+--------------------------------+----------+ > | 138 | root | localhost | NULL | Query > | 0 | init | SHOW FULL processlist > | 0.000 | > | 139 | hibernate_orm_test | localhost:52853 | hibernate_orm_test | Sleep > | 12 | | NULL > | 0.000 | > | 140 | hibernate_orm_test | localhost:52854 | hibernate_orm_test | Query > | 12 | Waiting for table metadata lock | drop table if exists TheEntity > | 0.000 | > > +-----+--------------------+-----------------+--------------------+---------+------+---------------------------------+--------------------------------+----------+ > > So we see 2 JDBC connections with process ids 139 and 140. You can see > 139 blocked waiting to be able to drop a table; specifically waiting on a > lock of some sort. > > However, `SHOW OPEN TABLES;` shows no tables in use and no tables locked. > > Unfortunately that is the extent of my "MySQL debugging skills". > > > On Fri, Jul 17, 2015 at 11:14 AM Steve Ebersole <steve(a)hibernate.org&gt; > wrote: > >> Process != test. Process here == connection. You have multiple. >> >> On Fri, Jul 17, 2015, 11:12 AM Sanne Grinovero <sanne(a)hibernate.org&gt; >> wrote: >> >>> On 17 July 2015 at 16:25, Steve Ebersole <steve(a)hibernate.org&gt; wrote: >>> > It would seem that this is a regression with ORM, I agree. However, >>> we do >>> > not know how recent of a regression. And that is the problem. We >>> have no >>> > idea how far back to look. Because the ORM CI jobs against MySQL, >>> > PostrgreSQL, Oracle, etc have not been running for many months. IIRC >>> even >>> > when they were running on the old CI servers we had problems with >>> those jobs >>> > that were not looked at for some time. >>> >>> Sure, same problem here: not run these tests for months, and you're >>> right I could do a quick check on a slightly older branch which is >>> using ORM 4.3 and I have the same problem. >>> >>> > As far as what you see, I am not convinced it is what you think it is >>> in >>> > terms of auto-commit for queries run outside of a transaction. This >>> part >>> > has not changed in 5.0. Assuming you are using the "built in >>> Connection >>> > pool", there is a setting that controls the auto-commit mode of the >>> > connections we create (hibernate.connection.autocommit). That setting >>> has >>> > always been there, and it has always defaulted to false. This has not >>> > changed. In fact the same exact code is used in 4.3 and 5.0 for >>> creating >>> > these Connections: >>> > >>> org.hibernate.engine.jdbc.connections.internal.BasicConnectionCreator. >>> The >>> > autoCommit value in there comes from the setting I just mentioned. The >>> > logical outcome is the same even before 4.3, just that it was done >>> > differently. So bottom line, if you are not seeing auto-commit on >>> these >>> > Connections now, you never were. Now it is possible that their are >>> some >>> > ancillary changes that cause the lack of auto-commit to now be an issue >>> > where it was not before. So first, I'd need to know whether the >>> Connection >>> > here is auto-commit or is not. From there we can start drilling down >>> on >>> > what might be a cause. But keep in mind too, that with hangs and >>> deadlocks >>> > the process that hangs is often not the culprit; often it is the >>> previous >>> > process that is holding locks that is the culprit. >>> >>> I understand, but in this specific case I can reproduce the issue by >>> running a single test. >>> Also that psql SQL I shared clearly shows which statement is causing it >>> to hang, >>> and I could fix the test by just wrapping that statement in a >>> transaction. >>> >>> > >>> > My case is actually different, at least on the surface. Mine happens >>> on >>> > dropping tables after the test is complete (for MySQL anyway). Which >>> got me >>> > to looking at schema export and friends and how they handle >>> connections. >>> > They handle connections quite differently. Again the same as they >>> used to, >>> > but I mean very different from how sessions manage connections. In >>> schema >>> > export the connections are always set to auto-commit mode at the start >>> of >>> > processing. Then they never do anything with auto-commit afterwards. >>> I am >>> > not sure this is right. >>> >>> Actually it sounds like you're having the same problem. I debugged the >>> PosgreSQL tests first, but I had the same on MySQL (as in the >>> stacktrace I shared), >>> and the test also hangs on schema drop. >>> The problem is that the schema can't be dropped because a previous >>> statement - from a session which is now correctly closed - is still >>> holding locks. >>> >>> Thanks, >>> Sanne >>> >>> > >>> > >>> > >>> > >>> > On Fri, Jul 17, 2015 at 3:51 AM Sanne Grinovero <sanne(a)hibernate.org&gt; >>> wrote: >>> >> >>> >> Thanks Steve, I wasn't sure if I had found a problem. >>> >> Still, it looks like the Hibernate Search tests are not benefiting >>> >> from auto-commit, or I wouldn't have this problem right? Our testing >>> >> configuration properties have been the same for years and I'm just >>> >> re-enabling tests which haven't been run for a while, so I suspect >>> >> something changed in ORM. >>> >> >>> >> I think I'd expect Hibernate ORM to automatically enable auto-commit >>> >> on operations not running within a transaction, to make sure all locks >>> >> are released when the Session is closed. >>> >> >>> >> Thanks, >>> >> Sanne >>> >> >>> >> On 17 July 2015 at 00:14, Steve Ebersole <steve(a)hibernate.org&gt; wrote: >>> >> > For what it is worth, I am running into the same problems on MySQL >>> for >>> >> > the >>> >> > ORM tests in terms of blocks on dropping tables. PGSQL at least >>> >> > finishes, >>> >> > although I run into 55 errors which I still need to triage. >>> >> > >>> >> > On Thu, Jul 16, 2015 at 5:59 PM Steve Ebersole <steve(a)hibernate.org >>> > >>> >> > wrote: >>> >> >> >>> >> >> There is really nothing to improve. What would you suggest as an >>> >> >> improvement? I mean from a ORM perspective, we have an ato-commit >>> >> >> Connection (presumably) and run a query. How on earth would we >>> know to >>> >> >> expect that the database is holding locks for that? >>> >> >> >>> >> >> As far as "requiring transactions" I have tried to fly that plane >>> >> >> multiple >>> >> >> times in the past. I think it is absolutely ridiculous to allow >>> these >>> >> >> auto-commit use cases. In fact I personally wanted to take it to >>> the >>> >> >> more >>> >> >> extreme case of always using JTA and simply mimicking "single >>> resource >>> >> >> JTA >>> >> >> transactions" internally in the "JDBC transaction" case. Gavin >>> and I >>> >> >> both >>> >> >> did in fact. But we got a lot of push back. >>> >> >> >>> >> >> So I am not sure exactly what the improvement here might be. What >>> is >>> >> >> it >>> >> >> you were thinking specifically? >>> >> >> >>> >> >> On Thu, Jul 16, 2015 at 5:36 PM Sanne Grinovero < >>> sanne(a)hibernate.org&gt; >>> >> >>> >> >> wrote: >>> >> >>> >>> >> >>> I finally got to re-enable MariaDB and PostgreSQL based tests for >>> >> >>> Hibernate Search - which has been running on H2 only for some >>> months - >>> >> >>> and had to debug a case of a single test hanging for a long time. >>> >> >>> >>> >> >>> Essentially it would block - for hours - on the >>> SessionFactory#close() >>> >> >>> method, attempting to drop the database schema with the following >>> >> >>> statement: >>> >> >>> > alter table AncientBook_alternativeTitles drop constraint >>> >> >>> FKn8hhkmhof1mdgc4oi77ccq989 >>> >> >>> >>> >> >>> Dumping threads I would get a very similar stack trace on both >>> >> >>> databases; initially I thought someone had copy/pasted a socket >>> >> >>> handling bug from one JDBC driver to the other ;-) >>> >> >>> >>> >> >>> The PostgreSQL testsuite hanging: >>> >> >>> >>> >> >>> "main" prio=10 tid=0x00007f0f40009000 nid=0x5f7c runnable >>> >> >>> [0x00007f0f48956000] >>> >> >>> java.lang.Thread.State: RUNNABLE >>> >> >>> at java.net.SocketInputStream.socketRead0(Native Method) >>> >> >>> at java.net.SocketInputStream.read(SocketInputStream.java:152) >>> >> >>> at java.net.SocketInputStream.read(SocketInputStream.java:122) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.core.VisibleBufferedInputStream.readMore(VisibleBufferedInputStream.java:145) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.core.VisibleBufferedInputStream.ensureBytes(VisibleBufferedInputStream.java:114) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.core.VisibleBufferedInputStream.read(VisibleBufferedInputStream.java:73) >>> >> >>> at org.postgresql.core.PGStream.ReceiveChar(PGStream.java:274) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.core.v3.QueryExecutorImpl.processResults(QueryExecutorImpl.java:1660) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.core.v3.QueryExecutorImpl.execute(QueryExecutorImpl.java:257) >>> >> >>> - locked <0x00000007c11e3860> (a >>> >> >>> org.postgresql.core.v3.QueryExecutorImpl) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.jdbc2.AbstractJdbc2Statement.execute(AbstractJdbc2Statement.java:500) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.jdbc2.AbstractJdbc2Statement.executeWithFlags(AbstractJdbc2Statement.java:374) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.postgresql.jdbc2.AbstractJdbc2Statement.executeUpdate(AbstractJdbc2Statement.java:302) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.tool.hbm2ddl.DatabaseExporter.export(DatabaseExporter.java:47) >>> >> >>> at >>> >> >>> >>> org.hibernate.tool.hbm2ddl.SchemaExport.perform(SchemaExport.java:476) >>> >> >>> at >>> >> >>> >>> org.hibernate.tool.hbm2ddl.SchemaExport.execute(SchemaExport.java:430) >>> >> >>> at >>> org.hibernate.tool.hbm2ddl.SchemaExport.drop(SchemaExport.java:375) >>> >> >>> at >>> org.hibernate.tool.hbm2ddl.SchemaExport.drop(SchemaExport.java:371) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.internal.SessionFactoryImpl.close(SessionFactoryImpl.java:1069) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.search.test.util.FullTextSessionBuilder.close(FullTextSessionBuilder.java:149) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.search.test.util.FullTextSessionBuilder$1.evaluate(FullTextSessionBuilder.java:248) >>> >> >>> at org.junit.rules.RunRules.evaluate(RunRules.java:20) >>> >> >>> at org.junit.runners.ParentRunner.run(ParentRunner.java:309) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.execute(JUnit4Provider.java:264) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.executeTestSet(JUnit4Provider.java:153) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.invoke(JUnit4Provider.java:124) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.invokeProviderInSameClassLoader(ForkedBooter.java:200) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.runSuitesInProcess(ForkedBooter.java:153) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.main(ForkedBooter.java:103) >>> >> >>> >>> >> >>> The MariaDB testsuite hanging: >>> >> >>> >>> >> >>> "main" prio=10 tid=0x00007f8ca0009000 nid=0x4043 runnable >>> >> >>> [0x00007f8ca5f5c000] >>> >> >>> java.lang.Thread.State: RUNNABLE >>> >> >>> at java.net.SocketInputStream.socketRead0(Native Method) >>> >> >>> at java.net.SocketInputStream.read(SocketInputStream.java:152) >>> >> >>> at java.net.SocketInputStream.read(SocketInputStream.java:122) >>> >> >>> at >>> >> >>> >>> >> >>> >>> com.mysql.jdbc.util.ReadAheadInputStream.fill(ReadAheadInputStream.java:114) >>> >> >>> at >>> >> >>> >>> >> >>> >>> com.mysql.jdbc.util.ReadAheadInputStream.readFromUnderlyingStreamIfNecessary(ReadAheadInputStream.java:161) >>> >> >>> at >>> >> >>> >>> >> >>> >>> com.mysql.jdbc.util.ReadAheadInputStream.read(ReadAheadInputStream.java:189) >>> >> >>> - locked <0x00000007c0baa518> (a >>> >> >>> com.mysql.jdbc.util.ReadAheadInputStream) >>> >> >>> at com.mysql.jdbc.MysqlIO.readFully(MysqlIO.java:2499) >>> >> >>> at com.mysql.jdbc.MysqlIO.reuseAndReadPacket(MysqlIO.java:2952) >>> >> >>> at com.mysql.jdbc.MysqlIO.reuseAndReadPacket(MysqlIO.java:2941) >>> >> >>> at com.mysql.jdbc.MysqlIO.checkErrorPacket(MysqlIO.java:3489) >>> >> >>> at com.mysql.jdbc.MysqlIO.sendCommand(MysqlIO.java:1959) >>> >> >>> at com.mysql.jdbc.MysqlIO.sqlQueryDirect(MysqlIO.java:2113) >>> >> >>> at com.mysql.jdbc.ConnectionImpl.execSQL(ConnectionImpl.java:2562) >>> >> >>> - locked <0x00000007c0baa850> (a java.lang.Object) >>> >> >>> at >>> com.mysql.jdbc.StatementImpl.executeUpdate(StatementImpl.java:1664) >>> >> >>> - locked <0x00000007c0baa850> (a java.lang.Object) >>> >> >>> at >>> com.mysql.jdbc.StatementImpl.executeUpdate(StatementImpl.java:1583) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.tool.hbm2ddl.DatabaseExporter.export(DatabaseExporter.java:47) >>> >> >>> at >>> >> >>> >>> org.hibernate.tool.hbm2ddl.SchemaExport.perform(SchemaExport.java:476) >>> >> >>> at >>> >> >>> >>> org.hibernate.tool.hbm2ddl.SchemaExport.execute(SchemaExport.java:430) >>> >> >>> at >>> org.hibernate.tool.hbm2ddl.SchemaExport.drop(SchemaExport.java:375) >>> >> >>> at >>> org.hibernate.tool.hbm2ddl.SchemaExport.drop(SchemaExport.java:371) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.internal.SessionFactoryImpl.close(SessionFactoryImpl.java:1069) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.search.test.util.FullTextSessionBuilder.close(FullTextSessionBuilder.java:149) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.hibernate.search.test.util.FullTextSessionBuilder$1.evaluate(FullTextSessionBuilder.java:248) >>> >> >>> at org.junit.rules.RunRules.evaluate(RunRules.java:20) >>> >> >>> at org.junit.runners.ParentRunner.run(ParentRunner.java:309) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.execute(JUnit4Provider.java:264) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.executeTestSet(JUnit4Provider.java:153) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.junit4.JUnit4Provider.invoke(JUnit4Provider.java:124) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.invokeProviderInSameClassLoader(ForkedBooter.java:200) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.runSuitesInProcess(ForkedBooter.java:153) >>> >> >>> at >>> >> >>> >>> >> >>> >>> org.apache.maven.surefire.booter.ForkedBooter.main(ForkedBooter.java:103) >>> >> >>> >>> >> >>> On Gail's suggestion I went looking for database level locks; I >>> hadn't >>> >> >>> thought of that as I assumed it would have timed out more >>> aggressively >>> >> >>> rather than have me wait for hours. >>> >> >>> It turns out she was right and the reason for blocking was a >>> simple >>> >> >>> "count *" query being run as a post-test assertion, which we had >>> >> >>> forgotten to wrap in an "open transaction & commit" statements >>> pair. >>> >> >>> The test assertion would be successful, but apparently it would >>> hold >>> >> >>> on the table lock beyond closing the Session and, then fail to >>> drop >>> >> >>> the database at the teardown of the test. >>> >> >>> >>> >> >>> I'm wondering if this is expected, or if there's something in the >>> new >>> >> >>> transaction handling code which could be improved? It took me >>> several >>> >> >>> hours to figure this out; maybe I'm just not using ORM as >>> frequently >>> >> >>> as once :) >>> >> >>> If it's this critical to have the transaction, maybe it should be >>> >> >>> mandatory? >>> >> >>> >>> >> >>> And as a memo for next time, this is the query to figure out >>> details >>> >> >>> about locks on our testing db: >>> >> >>> > SELECT * FROM pg_locks pl LEFT JOIN pg_stat_activity psa ON >>> pl.pid >>> >> >>> = psa.pid where datname = 'testingdb'; >>> >> >>> >>> >> >>> Thanks, >>> >> >>> Sanne >>> >> >>> _______________________________________________ >>> >> >>> hibernate-dev mailing list >>> >> >>> hibernate-dev(a)lists.jboss.org >>> >> >>> https://lists.jboss.org/mailman/listinfo/hibernate-dev >>> >> _______________________________________________ hibernate-dev mailing list hibernate-dev(a)lists.jboss.org https://lists.jboss.org/mailman/listinfo/hibernate-dev