[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Emmanuel Bernard commented on HSEARCH-421: ------------------------------------------ yes we need some kind of error handling processor concept in async works so that the user can do whatever is needed (send a message, log in a file, ignore, etc). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman updated HSEARCH-421: ------------------------------------------ Attachment: HibernateSearch-421-24032010.patch -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Sanne Grinovero updated HSEARCH-421: ------------------------------------ Attachment: HibernateSearch-421-24032010-FixedStyle.patch Amin's patch, just changed formatting to our conventions -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman commented on HSEARCH-421: ----------------------------------------------- Hi Please find attached a patch with the usage of the ErrorHandler. I've removed the Operation class as it was duplicating the LuceneWork so I've just used that. I think there may be comments on the usage of the error handler in the PerDPProcessor, just wasn't sure about getting the primary exception and subsequent failures. Would be grateful for your thoughts. The default implementation may need some more work which I'm guess you will want. Thanks P.S. I think the formatting should be more aligned to your convention. IntelliJ was doing some optimisation of the imports hence the inconsistencies. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Sanne Grinovero updated HSEARCH-421: ------------------------------------ Summary: Exceptions happening in backend are unnoticed (Amin Mohammed-Coleman) (was: Exceptions happening in backend are unnoticed) -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman commented on HSEARCH-421: ----------------------------------------------- Hi Sanne, Thanks for your comments. I've added some comments to explain some of the changes you mentioned: LogErrorHandler appending all to a single string to call log.error only once - so that it doesn't get mixed up with other message - I thought this would be case. I changed it this morning on the train. making sure the ErrorHandler isn't build during each work but reused - That makes sense. I will create a patch for documentation. Can I ask a quick question? Does the implementation in the PerDPProcessors look ok? I wasn't completely sure if I had captured what you and Emmanuel wanted. Thanks again! -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Sanne Grinovero commented on HSEARCH-421: ----------------------------------------- Another hint about the code needed in PerDPQueueProcessor: don't forget to wrap also the optimization operations and the indexWriter close. Imho you should accumulate work which was done in some collection and so in case of failure you can find out which work was performed and which wasn't, but you should avoid attempts to recover or ignore errors, we still want to cancel the current queue. Consider the ErrorContextBuilder is your own child, you're free to change it to better match the need to track work which was performed before the failure. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman commented on HSEARCH-421: ----------------------------------------------- Updated PerDPQueueProcessor to use error context. The ErrorContextBuilder now keeps a track of the work that completed and the total work. This way it can determine what operations worked and which failied. The LuceneErrorHandlingTest now is passing based on implementation (i've uncommented the assertions out...thanks Sanne for the testcase...slap on the wrist for me for not including one!). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Sanne Grinovero commented on HSEARCH-421: ----------------------------------------- Hi Amin, I see you're still loosing the fight with your ide about controlling the tabs/spaces :D The code is ok, I'm applying it with some more changes: * tabs * An additional ErrorHandler which rethrows the exceptions, this one is only meaningfull when backend is working in sync mode, so the factory sill throw a SearchException is it's configured with an async backend. * the LuceneBackendQueueProcessor won't flag the whole queue failed, we don't know in this context - it might propagate the error generated from the index worker which is already containing such information. * Added tests for the new ErrorHandler I'll wait for you to document the two configurable ErrorHandlers to close the issue (please read the keywords needed for configuration from SearchFactoryImpl and notice that the logging one is default for async, while propagating back the exception is the default for sync. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman updated HSEARCH-421: ------------------------------------------ Attachment: HibernateSearch-421-05042010.patch -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Sanne Grinovero commented on HSEARCH-421: ----------------------------------------- using either a fully qualified classname of an implementation or a shorthand name for provided implementations is common across all modules in Search. You can see this convention used for DirectoryProvider, Optimization strategies, Worker, backend processor, filter caching, batch backend, readerprovider, lock strategies, ... It's up to us to clearly explain the possible options in the documentation. Could you review your patch and remove all reformatting changes which are not related to this issue? The ide might be out of control :) but you could open the patches with a text editor and verify. thanks for all help -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman commented on HSEARCH-421: ----------------------------------------------- I've updated the patch so it should only have stuff related to the error handling. I applied the patch locally to make sure it was ok, the problem I kept getting was the following: //previous section <para /> <section> //error handling section //content </section> </section> //this section is from previous para tag </chapter> I've tried modifying the patch file to fix this but it isn't working. Sorry guys about this. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Amin Mohammed-Coleman commented on HSEARCH-421: ----------------------------------------------- Updated document to reflect the fact the there is no rethrow exceptions as an option. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-421?pag... ] Emmanuel Bernard updated HSEARCH-421: ------------------------------------- Issue Type: New Feature (was: Bug) Summary: Provide a callback on indexing exception (Amin Mohammed-Coleman) (was: Exceptions happening in backend are unnoticed (Amin Mohammed-Coleman)) -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira