[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324.patch Very nice! Looks like we needed all kinds of IW syncs? I noticed that in addition to TestStressIndexing2, TestNRTThreads was also failing. The attached patch fixes both by adding a sync on DW for deletes (and the update doc delete term). Time to add the RAM usage? Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324.patch, LUCENE-2324.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324.patch Ok, TestNRTThreads works after 10+ iterations. TestStressIndexing2 works most of the time however with enough iterations, eg, ant test-core -Dtestcase=TestStressIndexing2 -Dtests.iter=30 it fails. I think that deletes are sneaking in because we're not sync'ed on DW as we're flushing the DWPT. Ideally some assertions would pick this up. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324.patch, LUCENE-2324.patch, LUCENE-2324.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324.patch I removed ThreadState and DWPT now extends ReentrantLock. I added assertions such as delete terms assert numDocsInRAM 0; which fails. I think we should focus on cleaning up the locking, ie, do we need to use synchronized in DWPT? Perhaps everything should use the RL? Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: test.out Are you also merging trunk in as svn up yields a lot of updates. There are new test failures in: TestSnapshotDeletionPolicy Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: test.out Here's the latest test.out. There's a lot of these: {code} [junit] junit.framework.AssertionFailedError: IndexFileDeleter doesn't know about file _5.fdt [junit] at org.apache.lucene.util.LuceneTestCase$LuceneTestCaseRunner.runChild(LuceneTestCase.java:1156) [junit] at org.apache.lucene.util.LuceneTestCase$LuceneTestCaseRunner.runChild(LuceneTestCase.java:1088) [junit] at org.apache.lucene.index.IndexWriter.filesExist(IndexWriter.java:3273) [junit] at org.apache.lucene.index.IndexWriter.startCommit(IndexWriter.java:3321) [junit] at org.apache.lucene.index.IndexWriter.prepareCommit(IndexWriter.java:2339) [junit] at org.apache.lucene.index.IndexWriter.commitInternal(IndexWriter.java:2410) [junit] at org.apache.lucene.index.IndexWriter.closeInternal(IndexWriter.java:1083) [junit] at org.apache.lucene.index.IndexWriter.close(IndexWriter.java:1027) [junit] at org.apache.lucene.index.IndexWriter.close(IndexWriter.java:991) [junit] at org.apache.lucene.index.TestAddIndexes.testMergeAfterCopy(TestAddIndexes.java:432) {code} Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324-SMALL.patch I added a FlushPolicy class, deletes record the ram used to DWPT and DW. Recording to DW is for global ram used. The TIW OOM is still occurring however. The delete calls to FlushControl are gone, I'm not sure what's going to be left of it. bq. I'm not sure we should queue. I wonder how much this'd slow down the single threaded case? Yes, that's a good point. bq. We shouldn't do global waiting anymore - this is what's great about DWPT. However we'll have global waiting for the flush all threads case. I think that can move down to DW though. Or should it simply be a sync in/on IW? bq. I thought we don't have sequence IDs anymore? The seqid lock was there, however it was removed in the last update. I think we need to clearly document the various locks and sync points as right now it's not clear enough to prevent deadlock situations. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324-SMALL.patch RAM accounting is slightly improved, we had two variables in DW keeping track of it. The extraneous is removed, however we're still OOMing in: ant test-core -Dtestcase=TestIndexWriter -Dtestmethod=testIndexingThenDeleting Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324-SMALL.patch Same as the last patch, however default deletes is added to DW to which deletes are added to when there are no available DWPTs. On flush all threads, default deletes is applied to the last segment with no doc limit. TestIndexReaderReopen now passes. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: test.out Here's a new test.out, I'll look at TestCheckIndex which should probably work. IndexFileDeleter doesn't know about file seems odd. We're OOMing in TestIndexWriter because we're not flushing by RAM (eg, it currently defaults to return false). Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324-SMALL.patch I added an assertion showing the lack of DWPTs when delete is called. deleteTermNoWait (which skips flush control) is called in update doc and now deadlock doesn't occur when executing TestIndexWriter. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: test.out Here's ant test-core output. Looks like it's deadlocking in TestIndexWriter? There are some IR.reopen failures, a null pointer, and a delete count I'll look at. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324-SMALL.patch Small patch fixing the num deletes test null pointer. The TestIndexReaderReopen failure seems to have something to do with flushing deletes. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: LUCENE-2324-SMALL.patch, lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch, test.out See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Busch updated LUCENE-2324: -- Attachment: lucene-2324.patch Finally a new version of the patch! (Sorry for keeping you guys waiting...) It's not done yet, but it compiles (against realtime branch!) and 95% of the core test cases pass. Work done in addition to last patch: - Added DocumentsWriterPerThread - Reimplemented big parts of DocumentsWriter - Added DocumentsWriterThreadPool which is an extension point for different pool implementation. The default impl is the ThreadAffinityDocumentsWriterThreadPool, which does what the old code did (try to assign a DWPT always to the same thread). It should be easy now to add Document#getSourceID() and another pool that can assign threads based on the sourceID. - Initial implementation of sequenceIDs. Currently they're only used to keep track of deletes and not for e.g. NRT readers yet. - Lots of other changes here and there. TODOs: - Implement flush-by-ram logic - Implement logic to discard deletes from the deletes buffer - Finish sequenceID handling: IW#commit() and IW#close() should return ID of last flushed sequenceID - Maybe change delete logic: currently deletes are applied when a segment is flushed. Maybe we can keep it this way in the realtime-branch though, because that's most likely what we want to do once the RAM buffer is searchable and deletes are cheaper as they can then be done in-memory before flush - Fix unit tests (mostly exception handling and thread safety) - New test cases, e.g. for sequenceID testing - Simplify code: In some places I copied code around, which can probably be further simplified - I started removing some of the old setters/getters in IW which are not in IndexWriterConfig - need to finish that, or revert those changes and use a different patch - Fix nocommits - Performance testing I'm planning to commit this soon to the realtime branch, even though it's obviously not done yet. But it's a big patch and changes will be easier to track with an svn history. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: 4.0 Attachments: lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Busch updated LUCENE-2324: -- Fix Version/s: Realtime Branch (was: 4.0) Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: Realtime Branch Attachments: lucene-2324.patch, lucene-2324.patch, LUCENE-2324.patch See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Busch updated LUCENE-2324: -- Attachment: lucene-2324.patch The patch removes all *PerThread classes downstream of DocumentsWriter. This simplifies a lot of the flushing logic in the different consumers. The patch also removes FreqProxMergeState, because we don't have to interleave posting lists from different threads anymore of course. I really like these simplifications! There is still a lot to do: The changes in DocumentsWriter and IndexWriter are currently just experimental to make everything compile. Next I will introduce DocumentsWriterPerThread and implement the sequenceID logic (which was discussed here in earlier comments) and the new RAM management. I also want to go through the indexing chain once again - there are probably a few more things to clean up or simplify. The patch compiles and actually a surprising amount of tests pass. Only multi-threaded tests seem to fail, which is not very surprising, considering I removed all thread-handling logic from DocumentsWriter. :) So this patch isn't working yet - just wanted to post my current progress. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: 3.1 Attachments: lucene-2324.patch, LUCENE-2324.patch See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: https://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira - To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jason Rutherglen updated LUCENE-2324: - Attachment: LUCENE-2324.patch The patch is not committable. The basics are here. We can worry about max threads later. For now there's a doc writer per thread. I still don't fully understand remap deletes, so it's commented out. There's a flushedDocCount per DW, and a global one in IW. IWC is accessed directly where possible. When running TestIndexWriter, the DW resumeAllThreads assertion fails. If the total ram buffer size is too large, and the current thread's DW is the largest, it's flushed. The next test case can be in regards to the memory management. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: 3.1 Attachments: LUCENE-2324.patch See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Busch updated LUCENE-2324: -- Attachment: (was: lucene-2324-no-pooling.patch) Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: 3.1 See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org
[jira] Updated: (LUCENE-2324) Per thread DocumentsWriters that write their own private segments
[ https://issues.apache.org/jira/browse/LUCENE-2324?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Busch updated LUCENE-2324: -- Attachment: lucene-2324-no-pooling.patch All tests pass but I have to review if with the changes the memory consumption calculation still works correctly. Not sure if the junits test that? Also haven't done any performance testing yet. Per thread DocumentsWriters that write their own private segments - Key: LUCENE-2324 URL: https://issues.apache.org/jira/browse/LUCENE-2324 Project: Lucene - Java Issue Type: Improvement Components: Index Reporter: Michael Busch Assignee: Michael Busch Priority: Minor Fix For: 3.1 Attachments: lucene-2324-no-pooling.patch See LUCENE-2293 for motivation and more details. I'm copying here Mike's summary he posted on 2293: Change the approach for how we buffer in RAM to a more isolated approach, whereby IW has N fully independent RAM segments in-process and when a doc needs to be indexed it's added to one of them. Each segment would also write its own doc stores and normal segment merging (not the inefficient merge we now do on flush) would merge them. This should be a good simplification in the chain (eg maybe we can remove the *PerThread classes). The segments can flush independently, letting us make much better concurrent use of IO CPU. -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online. - To unsubscribe, e-mail: java-dev-unsubscr...@lucene.apache.org For additional commands, e-mail: java-dev-h...@lucene.apache.org