Re: [Hibernate] Timeout for tests
Ryan Campbell wrote: Can there be a timeout for the tests? I'm seeing tests hang against different databases. If there was a timeout for the junit tests, these hanging tests would be reported as failures. I ran the tests on MySQL, and the longest one was 255 seconds, so perhaps 500 seconds is long enough? Is that only one test, if yes, which one? I remember one of them running between 20-30 seconds but 4 minutes seems to be a real exception. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_id=7637alloc_id=16865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
RE: [Hibernate] Timeout for tests
There were at least two which were taking 200 seconds. I have lost the results I used to give you this total. The mysql and oracle runs are taking about 20 minutes each (total). timesten is taking about an hour. SQL server (jtds) and Sybase (jconnect) are both timing out completely because they fail to complete in under 3 hours. This could be a config problem, but the consistent problems I am having with long running tests make me wonder if it is some new test introducing some contention? -Original Message- From: Christian Bauer [mailto:[EMAIL PROTECTED] Sent: Monday, November 28, 2005 7:15 AM To: Ryan Campbell Cc: Hibernate development Subject: Re: [Hibernate] Timeout for tests Ryan Campbell wrote: Can there be a timeout for the tests? I'm seeing tests hang against different databases. If there was a timeout for the junit tests, these hanging tests would be reported as failures. I ran the tests on MySQL, and the longest one was 255 seconds, so perhaps 500 seconds is long enough? Is that only one test, if yes, which one? I remember one of them running between 20-30 seconds but 4 minutes seems to be a real exception. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
RE: [Hibernate] Timeout for tests
Also, there is the problem that Steve and I saw with the org.hibernate.test.legacy.PerformanceTest. It does not have a finally block for closing resources in case of error, and some databases will hang when the hibernate test case attempts to drop a table which has transactions still open against it. Perhaps this is the problem with jtds and Sybase? At any rate, the timeout would help diagnose this without having to reproduce manually. -Original Message- From: [EMAIL PROTECTED] [mailto:hibernate-devel- [EMAIL PROTECTED] On Behalf Of Ryan Campbell Sent: Monday, November 28, 2005 7:29 AM To: Christian Bauer Cc: Hibernate development Subject: RE: [Hibernate] Timeout for tests There were at least two which were taking 200 seconds. I have lost the results I used to give you this total. The mysql and oracle runs are taking about 20 minutes each (total). timesten is taking about an hour. SQL server (jtds) and Sybase (jconnect) are both timing out completely because they fail to complete in under 3 hours. This could be a config problem, but the consistent problems I am having with long running tests make me wonder if it is some new test introducing some contention? -Original Message- From: Christian Bauer [mailto:[EMAIL PROTECTED] Sent: Monday, November 28, 2005 7:15 AM To: Ryan Campbell Cc: Hibernate development Subject: Re: [Hibernate] Timeout for tests Ryan Campbell wrote: Can there be a timeout for the tests? I'm seeing tests hang against different databases. If there was a timeout for the junit tests, these hanging tests would be reported as failures. I ran the tests on MySQL, and the longest one was 255 seconds, so perhaps 500 seconds is long enough? Is that only one test, if yes, which one? I remember one of them running between 20-30 seconds but 4 minutes seems to be a real exception. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=ick ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
RE: [Hibernate] Timeout for tests
Can you give me the connection info to the various db instances? I'll try running these manually and see how long things are taking. Or have you tried this already? With Oracle for example, it is pretty important how the underlying database is setup in terms of how fast/slow the testsuite runs. In particular there is an option (the default) on the tablespace for whether or not to log operations. This logged information is then used in database recovery. But this logging really slows things down quite a bit. In my local environment, I created a dedicated tablespace where the hibernate tests occur and diable logging within this tablespace; this really sped up the testsuite (plus we don't really care about recovery of the schema here). -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ryan Campbell Sent: Monday, November 28, 2005 7:29 AM To: Christian Bauer Cc: Hibernate development Subject: RE: [Hibernate] Timeout for tests There were at least two which were taking 200 seconds. I have lost the results I used to give you this total. The mysql and oracle runs are taking about 20 minutes each (total). timesten is taking about an hour. SQL server (jtds) and Sybase (jconnect) are both timing out completely because they fail to complete in under 3 hours. This could be a config problem, but the consistent problems I am having with long running tests make me wonder if it is some new test introducing some contention? -Original Message- From: Christian Bauer [mailto:[EMAIL PROTECTED] Sent: Monday, November 28, 2005 7:15 AM To: Ryan Campbell Cc: Hibernate development Subject: Re: [Hibernate] Timeout for tests Ryan Campbell wrote: Can there be a timeout for the tests? I'm seeing tests hang against different databases. If there was a timeout for the junit tests, these hanging tests would be reported as failures. I ran the tests on MySQL, and the longest one was 255 seconds, so perhaps 500 seconds is long enough? Is that only one test, if yes, which one? I remember one of them running between 20-30 seconds but 4 minutes seems to be a real exception. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=ick ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
Re: [Hibernate] Timeout for tests
Ryan Campbell wrote: There were at least two which were taking 200 seconds. I have lost the results I used to give you this total. The mysql and oracle runs are taking about 20 minutes each (total). timesten is taking about an hour. SQL server (jtds) and Sybase (jconnect) are both timing out completely because they fail to complete in under 3 hours. This could be a config problem This is definitely a configuration problem. I can run all if these on my local machine in a matter of minutes. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_id=7637alloc_id=16865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
RE: [Hibernate] Timeout for tests
You might want to change whatever you are doing to just run the AllTests suite, which excludes a bunch of things. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Ryan Campbell Sent: Monday, November 28, 2005 8:39 AM To: Christian Bauer Cc: Hibernate development Subject: RE: [Hibernate] Timeout for tests Also, there is the problem that Steve and I saw with the org.hibernate.test.legacy.PerformanceTest. It does not have a finally block for closing resources in case of error, and some databases will hang when the hibernate test case attempts to drop a table which has transactions still open against it. Perhaps this is the problem with jtds and Sybase? At any rate, the timeout would help diagnose this without having to reproduce manually. -Original Message- From: [EMAIL PROTECTED] [mailto:hibernate-devel- [EMAIL PROTECTED] On Behalf Of Ryan Campbell Sent: Monday, November 28, 2005 7:29 AM To: Christian Bauer Cc: Hibernate development Subject: RE: [Hibernate] Timeout for tests There were at least two which were taking 200 seconds. I have lost the results I used to give you this total. The mysql and oracle runs are taking about 20 minutes each (total). timesten is taking about an hour. SQL server (jtds) and Sybase (jconnect) are both timing out completely because they fail to complete in under 3 hours. This could be a config problem, but the consistent problems I am having with long running tests make me wonder if it is some new test introducing some contention? -Original Message- From: Christian Bauer [mailto:[EMAIL PROTECTED] Sent: Monday, November 28, 2005 7:15 AM To: Ryan Campbell Cc: Hibernate development Subject: Re: [Hibernate] Timeout for tests Ryan Campbell wrote: Can there be a timeout for the tests? I'm seeing tests hang against different databases. If there was a timeout for the junit tests, these hanging tests would be reported as failures. I ran the tests on MySQL, and the longest one was 255 seconds, so perhaps 500 seconds is long enough? Is that only one test, if yes, which one? I remember one of them running between 20-30 seconds but 4 minutes seems to be a real exception. --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=ick ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=ick ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel --- This SF.net email is sponsored by: Splunk Inc. Do you grep through log files for problems? Stop! Download the new AJAX search engine that makes searching your log files as easy as surfing the web. DOWNLOAD SPLUNK! http://ads.osdn.com/?ad_idv37alloc_id865op=click ___ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
RE: [Hibernate] query statistics
So some things to consider then: A. You will only really, truly be able to get the SQL(s) after all parameters values are known (accounting for parameter-lists) B. Parameter-lists will again cause you problems here. This is because the original parameter name is no longer known by the time the query gets to the translator, thus it is not in the parameter-metadata handed to the QueryImpl. The parameter metadata contains the expanded list of names (i.e., originalName0_, originalName1_,...). From: Max Andersen Sent: Thursday, November 24, 2005 6:05 AM To: Steve Ebersole; Gavin King; Hibernate devel Subject: RES: [Hibernate] query statistics just a quick response. the priorities for the tools is: A. get the underlying sql(s) B. get the guessed types of the parameters C. get the locations of the parameters in HQL and SQL A and B is most important (and should be gettable without actually executing the query), C is not important since I can, at least for HQL, aproximate this. So dont let C hold a better solution back if that is the case. ;max De: [EMAIL PROTECTED] em nome de Steve Ebersole Enviada: qui 24-11-2005 03:53 Para: Gavin King; Hibernate devel Assunto: RE: [Hibernate] query statistics Well more I was thinking of delaying the param-list expansion until execution-time and handling it behind either the query-plan or the translators. There is an issue with the tools (or clients of Query in general) getting access to the actual sql to be performed. Essentially, it would not be available until after all parameter lists have been set. But afaict, this is the case regardless of which approach we choose here. So the other option would be to keep param list expansion in AbstractQueryImpl, but then pass both the source query and the expanded query to locate the query plan; the source query really only being used for statistics. Another issue here (though afaict this only affects tools) is proper resolution of named parameter locations corresponding to parameter-lists. This is due to the fact that, in the current setup, the translator only ever knows about the expanded query. -Original Message- From: Gavin King Sent: Monday, November 21, 2005 8:55 AM To: Steve Ebersole; 'Hibernate devel' Subject: RE: [Hibernate] query statistics Right. The only downside would be increased mem usage, I suppose. -Original Message- From: Steve Ebersole Sent: Monday, November 21, 2005 6:54 AM To: Gavin King; 'Hibernate devel' Subject: RE: [Hibernate] query statistics I guess we could make it so. This issue is that currently it does not know the non-expanded query. -Original Message- From: Gavin King Sent: Monday, November 21, 2005 8:51 AM To: Steve Ebersole; Hibernate devel Subject: RE: [Hibernate] query statistics Could the query plan cache both expanded and non-expanded SQL? -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Steve Ebersole Sent: Monday, November 21, 2005 6:49 AM To: Hibernate devel Subject: [Hibernate] query statistics In relation to http://opensource2.atlassian.com/projects/hibernate/browse/HHH-73 (Statistics for HQL queries use pre-processed query string): The original issue behind this case is now easy to fix because of the introduction of query plans. The query plans know about the query strings before any processing has occurred. So simply moving the stats API calls into the query plans will get around the mentioned issue. However, I brought up another point there regarding parameter-list expansion. Currently, this is done within the QueryImpls, such that the expansion has occurred before the query string even gets to query plans/translators. This leads to scenarios where the HQL: from Animal where description in (:descriptionList) Results in different statistics being tracked based on the number of parameters actually contained in the bound descriptionList. To me, for the same arguments as the original issue, it is non-intuitive to issue from Animal where description in (:descriptionList) as an HQL query, but to see things like from Animal where description in (:descriptionList0_, :descriptionList1_) in the statistics. The flip side of this is that correcting this behavior means delaying generation of the actual SQL until the QueryParameters are actually known. AFAICT, this really only effects the tools since the tools want to display the translated SQL within the console; however, the SQL may not be fully resolvable at that point in time if we start delaying the resolution of the SQL till execution. So I wanted to open this up for discussion. What makes the most sense here? --- This SF.Net email is sponsored by the JBoss Inc. Get Certified Today Register for a JBoss Training Course. Free Certification Exam for All Training Attendees Through End of 2005. For more
