[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: ConcatableAggregatedLogsProposal_v8.pdf Uploaded v8 doc based on the latest discussions. > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CombinedAggregatedLogsProposal_v6.pdf, CombinedAggregatedLogsProposal_v7.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, > ConcatableAggregatedLogsProposal_v4.pdf, > ConcatableAggregatedLogsProposal_v5.pdf, > ConcatableAggregatedLogsProposal_v8.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: CombinedAggregatedLogsProposal_v7.pdf I’ve uploaded a v7 doc which uses the new log aggregation status. Both the v6 and the v7 designs have advantages and disadvantages over the other, so I’ve created the below list. ||v6 design||v7 design|| |depends on ZooKeeper|*no external dependencies*| |requires synchronizing all NMs|*requires no extra synchronization*| |*can combine even if some aggregated logs fail*|requires that all aggregated logs for an app succeed in order to combine them| |*implicitly distributes the HDFS load across the cluster for better balance*|HDFS load concentrated on RM’s datanode or all writes are remote| |overall is more complicated|*overall is simpler*| (bold is better) Please let me know if anyone has any feedback on the v7 design and which design is preferred. > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CombinedAggregatedLogsProposal_v6.pdf, CombinedAggregatedLogsProposal_v7.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, > ConcatableAggregatedLogsProposal_v4.pdf, > ConcatableAggregatedLogsProposal_v5.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: CombinedAggregatedLogsProposal_v6.pdf I've uploaded a v6 design (CombinedAggregatedLogsProposal_v6.pdf) based on offline discussion with Vinod and Karthik. It's mostly the original design where logs are appended (concat is not used), but we now have a Service that handles the appending and we're going to store the list of applications that need to be appended so recovery will work. As long as the NM comes back up, all logs should eventually be combined now. [~vinodkv], can you take a look at the document? > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CombinedAggregatedLogsProposal_v6.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, > ConcatableAggregatedLogsProposal_v4.pdf, > ConcatableAggregatedLogsProposal_v5.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: ConcatableAggregatedLogsProposal_v5.pdf I've uploaded a v5 doc which address those changes. I also clarified a few other things in there too. > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, > ConcatableAggregatedLogsProposal_v4.pdf, > ConcatableAggregatedLogsProposal_v5.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: ConcatableAggregatedLogsProposal_v4.pdf I've just uploaded ConcatableAggregatedLogsProposal_v4.pdf, with an updated design that uses a slightly modified version of the CombinedAggregatedLogFormat (now ConcatableAggregatedLogFormat) I already wrote and would use HDFS concat to combine the files. [~zjshen], [~kasha], and [~vinodkv], can you take a look at it? > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, > ConcatableAggregatedLogsProposal_v4.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Attachment: CombinedAggregatedLogsProposal_v3.pdf I've just uploaded CombinedAggregatedLogsProposal_v3.pdf, which has some minor updates. > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CombinedAggregatedLogsProposal_v3.pdf, > CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (YARN-2942) Aggregated Log Files should be combined
[ https://issues.apache.org/jira/browse/YARN-2942?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Robert Kanter updated YARN-2942: Summary: Aggregated Log Files should be combined (was: Aggregated Log Files should be compacted) > Aggregated Log Files should be combined > --- > > Key: YARN-2942 > URL: https://issues.apache.org/jira/browse/YARN-2942 > Project: Hadoop YARN > Issue Type: New Feature >Affects Versions: 2.6.0 >Reporter: Robert Kanter >Assignee: Robert Kanter > Attachments: CompactedAggregatedLogsProposal_v1.pdf, > CompactedAggregatedLogsProposal_v2.pdf, YARN-2942-preliminary.001.patch, > YARN-2942-preliminary.002.patch, YARN-2942.001.patch, YARN-2942.002.patch, > YARN-2942.003.patch > > > Turning on log aggregation allows users to easily store container logs in > HDFS and subsequently view them in the YARN web UIs from a central place. > Currently, there is a separate log file for each Node Manager. This can be a > problem for HDFS if you have a cluster with many nodes as you’ll slowly start > accumulating many (possibly small) files per YARN application. The current > “solution” for this problem is to configure YARN (actually the JHS) to > automatically delete these files after some amount of time. > We should improve this by compacting the per-node aggregated log files into > one log file per application. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
