[jira] [Commented] (HBASE-6028) Implement a cancel for in-progress compactions
[ https://issues.apache.org/jira/browse/HBASE-6028?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13279277#comment-13279277 ] Derek Wollenstein commented on HBASE-6028: -- @Stack -- I just looked at the current version -- it looks like the current check for this is in StoreFile.Writer.compactStore(final CollectionStoreFile filesToCompact, final boolean majorCompaction, final long maxId). // check periodically to see if a system stop is requested if (Store.closeCheckInterval 0) { bytesWritten += kv.getLength(); if (bytesWritten Store.closeCheckInterval) { bytesWritten = 0; if (!this.region.areWritesEnabled()) { writer.close(); fs.delete(writer.getPath(), false); throw new InterruptedIOException( Aborting compaction of store + this + in region + this.region + because user requested stop.); } } } So we could go ahead and actually put an explicit compactions enabled feature. This would also help if I wanted to later on add a feature to put in a time window for compactions. Implement a cancel for in-progress compactions -- Key: HBASE-6028 URL: https://issues.apache.org/jira/browse/HBASE-6028 Project: HBase Issue Type: Bug Components: regionserver Reporter: Derek Wollenstein Priority: Minor Labels: compaction, operations, regionserver Depending on current server load, it can be extremely expensive to run periodic minor / major compactions. It would be helpful to have a feature where a user could use the shell or a client tool to explicitly cancel an in-progress compactions. This would allow a system to recover when too many regions became eligible for compactions at once -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Status: Open (was: Patch Available) New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: (was: HBASE-5920.patch) New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: HBASE-5920-0.92.1.patch Creating a new version of the patch with svn New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Release Note: This patch makes the following changes: 1) Trace-level logging whenever a compaction is requested 2) Debug-level logging whenever compaction is changed 3) If a user requests a major compaction, this will stay a major compaction even if it violated max files (easy to take this part out) 3a) If a user-initiates a max compaction that requires too many files to be compacted, this will log an error. 4) Migrates utility functions from HBaseTestCase (Deprecated?) to HBaseTestingUtility to ease testing compaction behavior in TestCompaction Status: Patch Available (was: Open) Trying one more time -- I was just using diff rather than svn diff last time. New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: HBASE-5920-0.92.1-1.patch changing location again New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13276570#comment-13276570 ] Derek Wollenstein commented on HBASE-5920: -- I just noticed that this seems to be a jenkins error(patches seem to have been failing for several days?) I'll reach out via IRC / etc to see what's wrong here At revision 1339047. HBASE-5920 patch is being downloaded at Wed May 16 08:01:44 UTC 2012 from http://issues.apache.org/jira/secure/attachment/12527581/HBASE-5920-0.92.1-1.patch cp: cannot stat `/home/jenkins/buildSupport/lib/*': No such file or directory New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA,
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: HBASE-5920-trunk.patch Thanks @zhihong - I've re-created my patch against trunk. I've also removed some of the refactoring of unit tests since it looks like HBaseTestCase is continuing to see expanded use (and I don't want to mix a test refactoring with a compaction change) New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch, HBASE-5920-trunk.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: (was: HBASE-5920-trunk.patch) New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch, HBASE-5920-trunk.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: HBASE-5920-trunk.patch That was odd -- the patch in the jenkins log was empty -- trying to upload again New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch, HBASE-5920-trunk.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13276973#comment-13276973 ] Derek Wollenstein commented on HBASE-5920: -- Is there anything I can be doing better on my side? I have been running tests on my side iwhtout running into these issues, but I'm clearly doing something wrong New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920-0.92.1-1.patch, HBASE-5920-0.92.1.patch, HBASE-5920-trunk.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Created] (HBASE-6028) Implement a cancel for in-progress compactions
Derek Wollenstein created HBASE-6028: Summary: Implement a cancel for in-progress compactions Key: HBASE-6028 URL: https://issues.apache.org/jira/browse/HBASE-6028 Project: HBase Issue Type: Bug Components: regionserver Reporter: Derek Wollenstein Priority: Minor Depending on current server load, it can be extremely expensive to run periodic minor / major compactions. It would be helpful to have a feature where a user could use the shell or a client tool to explicitly cancel an in-progress compactions. This would allow a system to recover when too many regions became eligible for compactions at once -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Status: Patch Available (was: Open) This patch makes the following changes: 1) Trace-level logging whenever a compaction is requested 2) Debug-level logging whenever compaction is changed 3) If a user requests a major compaction, this will stay a major compaction even if it violated max files (easy to take this part out) 3a) If a user-initiates a max compaction that requires too many files to be compacted, this will log an error. 4) Migrates utility functions from HBaseTestCase (Deprecated?) to HBaseTestingUtility to ease testing compaction behavior in TestCompaction New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators:
[jira] [Updated] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Derek Wollenstein updated HBASE-5920: - Attachment: HBASE-5920.patch I'm not sure if this is being done correctly, but I've provided a patch that seems to do the right thing here. It includes a large change area because of the refactoring of HBaseTestCase utilities into HBaseTestingUtility. New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction Attachments: HBASE-5920.patch There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13272097#comment-13272097 ] Derek Wollenstein commented on HBASE-5920: -- I don't have a patch right now, but let me try to get something ready by early next week -- this should be relatively simple to push through. Honestly, I could even deal with the delay if it wasn't just something so silent from the perspective of the client. I only verified that my call really worked by turning on RPC Debug logging. New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Created] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
Derek Wollenstein created HBASE-5920: Summary: New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for user enabled compactions. I'd also like to see some override that says user triggered major compaction always occurs, but maybe that's a bad idea for other reasons. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-5920) New Compactions Logic can silently prevent user-initiated compactions from occurring
[ https://issues.apache.org/jira/browse/HBASE-5920?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13266980#comment-13266980 ] Derek Wollenstein commented on HBASE-5920: -- Nicolas -- Regarding the queue size -- no question. This is the result of aggressively disabling compactions with the intend of manually triggering them later, and then waiting two weeks. I was just upset when all the manually triggered compactions went into a black hole. Before modifying all these throttling settings, I was getting no information whatsoever in logs. the only reason I even knew the RPC worked was that when I turned on IPC debug logging I got a message saying that the request was serviced. All regions definitely were congested: this is because automatic compactions hadn't been running, and we weren't able to trigger manual compactions. We were going to be running these 1 region at a time, but since we had to restart the regionservers to get new settings (correct me if this is wrong), we ended up enqueueing all of the 2-week-delayed compactions at once. New Compactions Logic can silently prevent user-initiated compactions from occurring Key: HBASE-5920 URL: https://issues.apache.org/jira/browse/HBASE-5920 Project: HBase Issue Type: Bug Components: client, regionserver Affects Versions: 0.92.1 Reporter: Derek Wollenstein Priority: Minor Labels: compaction There seem to be some tuning settings in which manually triggered major compactions will do nothing, including loggic From Store.java in the function ListStoreFile compactSelection(ListStoreFile candidates) When a user manually triggers a compaction, this follows the same logic as a normal compaction check. when a user manually triggers a major compaction, something similar happens. Putting this all together: 1. If a user triggers a major compaction, this is checked against a max files threshold (hbase.hstore.compaction.max). If the number of storefiles to compact is max files, then we downgrade to a minor compaction 2. If we are in a minor compaction, we do the following checks: a. If the file is less than a minimum size (hbase.hstore.compaction.min.size) we automatically include it b. Otherwise, we check how the size compares to the next largest size. based on hbase.hstore.compaction.ratio. c. If the number of files included is less than a minimum count (hbase.hstore.compaction.min) then don't compact. In many of the exit strategies, we aren't seeing an error message. The net-net of this is that if we have a mix of very large and very small files, we may end up having too many files to do a major compact, but too few files to do a minor compact. I'm trying to go through and see if I'm understanding things correctly, but this seems like the bug To put it another way 2012-05-02 20:09:36,389 DEBUG org.apache.hadoop.hbase.regionserver.CompactSplitThread: Large Compaction requested: regionName=str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e., store Name=c, fileCount=15, fileSize=1.5g (20.2k, 362.5m, 155.3k, 3.0m, 30.7k, 361.2m, 6.9m, 4.7m, 14.7k, 363.4m, 30.9m, 3.2m, 7.3k, 362.9m, 23.5m), priority=-9, time=3175046817624398; Because: Recursive enqueue; compaction_queue=(59:0), split_queue=0 When we had a minimum compaction size of 128M, and default settings for hbase.hstore.compaction.min,hbase.hstore.compaction.max,hbase.hstore.compaction.ratio, we were not getting a compaction to run even if we ran major_compact 'str,44594594594594592,1334939064521.f7aed25b55d4d7988af763bede9ce74e.' from the ruby shell. Note that we had many tiny regions (20k, 155k, 3m, 30k,..) and several large regions (362.5m,361.2m,363.4m,362.9m). I think the bimodal nature of the sizes prevented us from doing a compaction. I'm not 100% sure where this errored out because when I manually triggered a compaction, I did not see ' // if we don't have enough files to compact, just wait if (filesToCompact.size() this.minFilesToCompact) { if (LOG.isDebugEnabled()) { LOG.debug(Skipped compaction of + this.storeNameStr + . Only + (end - start) + file(s) of size + StringUtils.humanReadableInt(totalSize) + have met compaction criteria.); } ' being printed in the logs (and I know DEBUG logging was enabled because I saw this elsewhere). I'd be happy with better error messages when we decide not to compact for