[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16963458#comment-16963458 ] Tsz-wo Sze commented on HDDS-2331: -- Filed HDDS-2386 for incremental buffering. > Shall we resolve it, ... +1, let's resolve this. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16962669#comment-16962669 ] Jitendra Nath Pandey commented on HDDS-2331: Shall we resolve it, given RATIS-726 is committed, and HDDS-2375 tracks the changes to optimize buffer allocation? > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16961613#comment-16961613 ] Tsz-wo Sze commented on HDDS-2331: -- Created HDDS-2375 to refactor the code. Will implement chunk buffer using a list of smaller buffers, which are allocated only if needed, after that. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16956020#comment-16956020 ] Attila Doroszlai commented on HDDS-2331: Thanks [~shashikant] for the tip about setting chunk size to 1MB. By doing so, writing 5K keys of size 1MB succeeds, even with 512MB max. heap. Taking a dump at around 4K keys shows similar ~170 ContainerCommandRequestMessage instances. (And yes, only half of the instances (writeChunk) have data buffer, the other half is empty (putBlock)). > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955983#comment-16955983 ] Tsz-wo Sze commented on HDDS-2331: -- Chunk size and array size may not be necessarily the same. Chunk size (e.g. 16MB) is a larger value than array size (e.g. 1MB). A chunk could use a list of arrays as the buffer. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955856#comment-16955856 ] Shashikant Banerjee commented on HDDS-2331: --- One more point to add here is, as per the cmd, freon is run with a single thread itself, which i assume will write only 1 key at a time. For a 1MB key, we should have only at max 2 containerCommadRequestMsgs(one for write chunk and one for putBlock) and these objects should disappear/garbage collected after the response is received. At any point of time, it should not hence show up 150+ containerCommandRequestObjects lying around. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955814#comment-16955814 ] Shashikant Banerjee commented on HDDS-2331: --- In Ozone, by default the buffer size is equal to the chunk size(16 MB default). Once a write call happens, a buffer is allocated and data is being written just into the buffer till it gets full/flush/close and then pushed to datanode and the buffer gets only released when watchForCommit call for the respective putBlock call log Index succeeds successfully. So until and unless, watchForCommit call gets acknowledged by Ratis, we keep holding onto the buffer so that, in case the ratis request fails, we have the user data cached in the client buffer which can be written over to the next block. We have had multiple discussions around this on reducing the default buffer size and implement a true streaming client, but this is still under consideration. [~adoroszlai], for your test, you can try changing the default chunk size to say 1 MB and see if it works well. It might also be possible that buffer release handling got broken with some changes introduced which need to be verified. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Assignee: Shashikant Banerjee >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955548#comment-16955548 ] Attila Doroszlai commented on HDDS-2331: Thanks for checking, [~szetszwo]. I probably should have said the bug is triggered by HDDS-2169, not caused by it. I agree, 16MB buffer is overkill for smaller keys. I tried to change it to match actual data length, but it's not trivial (causes other errors). > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955013#comment-16955013 ] Tsz-wo Sze commented on HDDS-2331: -- [~adoroszlai], thanks for discovering the problem. The finding indeed shows that HDDS-2169 is working well -- it is definitely able to avoid buffer copying. :) It also shows that Ozone probably needs a better buffer management. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16955011#comment-16955011 ] Tsz-wo Sze commented on HDDS-2331: -- >From https://issues.apache.org/jira/secure/attachment/12983374/profiler.png , >it seems that Ozone client creates a 16MB byte[] as a buffer to store an 1MB >data and then uses it to create a ByteString so that the 16MB byte[] is >referenced in the ByteString. Before HDDS-2169, the proto is *copied* to a >ByteString so that the 16MB byte[] buffer can be released. HDDS-2169 avoids >the copying and the buffer remains in the ByteString. In order to fix the problem, we may either (1) avoid creating a 16MB byte[] in the first place (for example, we may gradually create 1MB byte[] instead) or (2) give up on avoiding the buffer copying. It seems (1) is a better solution. BTW, please feel free to revert HDDS-2169 if it is desirable. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16954800#comment-16954800 ] Attila Doroszlai commented on HDDS-2331: Thanks [~aengineer]. I can workaround it by locally reverting the change. So if noone else complains I'm not rushing any revert in the repo. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16954792#comment-16954792 ] Anu Engineer commented on HDDS-2331: [~adoroszlai] Would you like me to revert this commit for now? > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. > Steps to reproduce: > # Start Ozone cluster with 1 datanode > # Start Freon (5K keys of size 1MB) > Result: OOME after a few hundred keys > {noformat} > $ cd hadoop-ozone/dist/target/ozone-0.5.0-SNAPSHOT/compose/ozone > $ docker-compose up -d > $ docker-compose exec scm bash > $ export HADOOP_OPTS='-XX:+HeapDumpOnOutOfMemoryError' > $ ozone freon rk --numOfThreads 1 --numOfVolumes 1 --numOfBuckets 1 > --replicationType RATIS --factor ONE --keySize 1048576 --numOfKeys 5120 > --bufferSize 65536 > ... > java.lang.OutOfMemoryError: Java heap space > Dumping heap to java_pid289.hprof ... > Heap dump file created [1456141975 bytes in 7.760 secs] > {noformat} -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
[jira] [Commented] (HDDS-2331) Client OOME due to buffer retention
[ https://issues.apache.org/jira/browse/HDDS-2331?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16954527#comment-16954527 ] Attila Doroszlai commented on HDDS-2331: Hi [~szetszwo], can you please check this issue? It seems to be caused by HDDS-2169, as reverting that [change|https://github.com/apache/hadoop-ozone/commit/e70ea7b6] allows Freon to continue creating keys. > Client OOME due to buffer retention > --- > > Key: HDDS-2331 > URL: https://issues.apache.org/jira/browse/HDDS-2331 > Project: Hadoop Distributed Data Store > Issue Type: Bug > Components: Ozone Client >Affects Versions: 0.5.0 >Reporter: Attila Doroszlai >Priority: Critical > Attachments: profiler.png > > > Freon random key generator exhausts default heap after just few hundred 1MB > keys. Heap dump on OOME reveals 150+ instances of > {{ContainerCommandRequestMessage}}, each with 16MB {{byte[]}}. -- This message was sent by Atlassian Jira (v8.3.4#803005) - To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org