[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13791525#comment-13791525
]
Hudson commented on YARN-1284:
--
FAILURE: Integrated in Hadoop-Mapreduce-trunk #1574 (See
[https://builds.apache.org/job/Hadoop-Mapreduce-trunk/1574/])
Amending yarn CHANGES.txt moving YARN-1284 to 2.2.1 (tucu:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530716)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.2.1
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13791485#comment-13791485
]
Hudson commented on YARN-1284:
--
FAILURE: Integrated in Hadoop-Hdfs-trunk #1548 (See
[https://builds.apache.org/job/Hadoop-Hdfs-trunk/1548/])
Amending yarn CHANGES.txt moving YARN-1284 to 2.2.1 (tucu:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530716)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.2.1
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13791384#comment-13791384
]
Hudson commented on YARN-1284:
--
SUCCESS: Integrated in Hadoop-Yarn-trunk #358 (See
[https://builds.apache.org/job/Hadoop-Yarn-trunk/358/])
Amending yarn CHANGES.txt moving YARN-1284 to 2.2.1 (tucu:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530716)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.2.1
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790628#comment-13790628
]
Hudson commented on YARN-1284:
--
SUCCESS: Integrated in Hadoop-trunk-Commit #4574 (See
[https://builds.apache.org/job/Hadoop-trunk-Commit/4574/])
Amending yarn CHANGES.txt moving YARN-1284 to 2.2.1 (tucu:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530716)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.2.1
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790347#comment-13790347
]
Hudson commented on YARN-1284:
--
SUCCESS: Integrated in Hadoop-Hdfs-trunk #1547 (See
[https://builds.apache.org/job/Hadoop-Hdfs-trunk/1547/])
Add missing file TestCgroupsLCEResourcesHandler for YARN-1284. (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530493)
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/util/TestCgroupsLCEResourcesHandler.java
YARN-1284. LCE: Race condition leaves dangling cgroups entries for killed
containers. (Alejandro Abdelnur via Sandy Ryza) (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530492)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/util/CgroupsLCEResourcesHandler.java
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.3.0
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790320#comment-13790320
]
Hudson commented on YARN-1284:
--
FAILURE: Integrated in Hadoop-Mapreduce-trunk #1573 (See
[https://builds.apache.org/job/Hadoop-Mapreduce-trunk/1573/])
Add missing file TestCgroupsLCEResourcesHandler for YARN-1284. (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530493)
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/util/TestCgroupsLCEResourcesHandler.java
YARN-1284. LCE: Race condition leaves dangling cgroups entries for killed
containers. (Alejandro Abdelnur via Sandy Ryza) (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530492)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/util/CgroupsLCEResourcesHandler.java
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.3.0
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790243#comment-13790243
]
Hudson commented on YARN-1284:
--
SUCCESS: Integrated in Hadoop-Yarn-trunk #357 (See
[https://builds.apache.org/job/Hadoop-Yarn-trunk/357/])
Add missing file TestCgroupsLCEResourcesHandler for YARN-1284. (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530493)
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/util/TestCgroupsLCEResourcesHandler.java
YARN-1284. LCE: Race condition leaves dangling cgroups entries for killed
containers. (Alejandro Abdelnur via Sandy Ryza) (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530492)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/util/CgroupsLCEResourcesHandler.java
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.3.0
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790036#comment-13790036
]
Hudson commented on YARN-1284:
--
SUCCESS: Integrated in Hadoop-trunk-Commit #4568 (See
[https://builds.apache.org/job/Hadoop-trunk-Commit/4568/])
Add missing file TestCgroupsLCEResourcesHandler for YARN-1284. (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530493)
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/util/TestCgroupsLCEResourcesHandler.java
YARN-1284. LCE: Race condition leaves dangling cgroups entries for killed
containers. (Alejandro Abdelnur via Sandy Ryza) (sandy:
http://svn.apache.org/viewcvs.cgi/?root=Apache-SVN&view=rev&rev=1530492)
* /hadoop/common/trunk/hadoop-yarn-project/CHANGES.txt
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java
*
/hadoop/common/trunk/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/util/CgroupsLCEResourcesHandler.java
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Fix For: 2.3.0
>
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13790020#comment-13790020
]
Sandy Ryza commented on YARN-1284:
--
+1
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789995#comment-13789995
]
Hadoop QA commented on YARN-1284:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12607500/YARN-1284.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 1 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:green}+1 findbugs{color}. The patch does not introduce any new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/2150//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/2150//console
This message is automatically generated.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789978#comment-13789978
]
Alejandro Abdelnur commented on YARN-1284:
--
For the record, I've spend a couple hours trying an alternate approach
suggested by [~rvs] while chatting offline about this. His suggestion was to
initialize a trash cgroup next to the containers cgroups and when a container
is cleanup transition the /tasks to the trash/tasks, doing the
equivalent of a {{cat /tasks >> trash/tasks}}. Tried doing that but
it seems some of the Java IO native calls make a system call which is not
supported by the cgroups filesystem implementation and I was getting the
following stack trace:
{code}
java.io.IOException: Argument list too long
java.io.IOException: Argument list too long
at java.io.FileOutputStream.writeBytes(Native Method)
at java.io.FileOutputStream.write(FileOutputStream.java:318)
at org.apache.hadoop.io.IOUtils.copyBytes(IOUtils.java:80)
...
{code}
Given this, beside that I didn't get it to work properly, I would not be
comfortable doing this as this may behave different in different Linux versions.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789977#comment-13789977
]
Hadoop QA commented on YARN-1284:
-
{color:red}-1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12607499/YARN-1284.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 1 new
or modified test files.
{color:red}-1 javac{color:red}. The patch appears to cause the build to
fail.
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/2149//console
This message is automatically generated.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch,
> YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789864#comment-13789864
]
Sandy Ryza commented on YARN-1284:
--
Oh, and also:
{code{
+if (! new File(cgroupPath).delete()) {
+ LOG.warn("Unable to delete cgroup at: " + cgroupPath +", tried to delete
for " +
+ deleteCgroupTimeout + "ms");
}
{code}
If the file was already deleted, delete() will return false and we'll log the
warning even though nothing went wrong. Instead, we should just check "if
(!deleted)".
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789861#comment-13789861
]
Sandy Ryza commented on YARN-1284:
--
A few nits. Otherwise LGTM.
{code}
+ //package private for testing purposes
+ private long deleteCgroupTimeout;
+ Clock clock;
{code}
Comment should go before the second variable. Also there should be a space
after the "//".
{code}
+ //visible for testing
{code}
Should the VisibleForTesting annotation be used? This is in two places.
{code}
+LOG.debug("deleteCgroup: " + cgroupPath);
{code}
Should be surrounded by if (LOG.isDebugEnabled())
{code}
+//file exists
{code}
Space after "//"?
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789769#comment-13789769
]
Alejandro Abdelnur commented on YARN-1284:
--
tested in a cluster using cgroups and works as expected, both the delete and
the timeouts.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789429#comment-13789429
]
Hadoop QA commented on YARN-1284:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12607388/YARN-1284.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 1 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:green}+1 findbugs{color}. The patch does not introduce any new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/2146//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/2146//console
This message is automatically generated.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789355#comment-13789355
]
Hadoop QA commented on YARN-1284:
-
{color:red}-1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12607374/YARN-1284.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 1 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:red}-1 findbugs{color}. The patch appears to introduce 2 new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/2145//testReport/
Findbugs warnings:
https://builds.apache.org/job/PreCommit-YARN-Build/2145//artifact/trunk/patchprocess/newPatchFindbugsWarningshadoop-yarn-server-nodemanager.html
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/2145//console
This message is automatically generated.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch, YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
[jira] [Commented] (YARN-1284) LCE: Race condition leaves dangling cgroups entries for killed containers
[
https://issues.apache.org/jira/browse/YARN-1284?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13789262#comment-13789262
]
Hadoop QA commented on YARN-1284:
-
{color:red}-1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12607362/YARN-1284.patch
against trunk revision .
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 1 new
or modified test files.
{color:green}+1 javac{color}. The applied patch does not increase the
total number of javac compiler warnings.
{color:green}+1 javadoc{color}. The javadoc tool did not generate any
warning messages.
{color:green}+1 eclipse:eclipse{color}. The patch built with
eclipse:eclipse.
{color:red}-1 findbugs{color}. The patch appears to introduce 1 new
Findbugs (version 1.3.9) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:green}+1 core tests{color}. The patch passed unit tests in
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/2144//testReport/
Findbugs warnings:
https://builds.apache.org/job/PreCommit-YARN-Build/2144//artifact/trunk/patchprocess/newPatchFindbugsWarningshadoop-yarn-server-nodemanager.html
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/2144//console
This message is automatically generated.
> LCE: Race condition leaves dangling cgroups entries for killed containers
> -
>
> Key: YARN-1284
> URL: https://issues.apache.org/jira/browse/YARN-1284
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
>Affects Versions: 2.2.0
>Reporter: Alejandro Abdelnur
>Assignee: Alejandro Abdelnur
>Priority: Blocker
> Attachments: YARN-1284.patch
>
>
> When LCE & cgroups are enabled, when a container is is killed (in this case
> by its owning AM, an MRAM) it seems to be a race condition at OS level when
> doing a SIGTERM/SIGKILL and when the OS does all necessary cleanup.
> LCE code, after sending the SIGTERM/SIGKILL and getting the exitcode,
> immediately attempts to clean up the cgroups entry for the container. But
> this is failing with an error like:
> {code}
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.LinuxContainerExecutor: Exit code
> from container container_1381179532433_0016_01_11 is : 143
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.containermanager.container.Container:
> Processing container_1381179532433_0016_01_11 of type
> UPDATE_DIAGNOSTICS_MSG
> 2013-10-07 15:21:24,359 DEBUG
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> deleteCgroup:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> 2013-10-07 15:21:24,359 WARN
> org.apache.hadoop.yarn.server.nodemanager.util.CgroupsLCEResourcesHandler:
> Unable to delete cgroup at:
> /run/cgroups/cpu/hadoop-yarn/container_1381179532433_0016_01_11
> {code}
> CgroupsLCEResourcesHandler.clearLimits() has logic to wait for 500 ms for AM
> containers to avoid this problem. it seems this should be done for all
> containers.
> Still, waiting for extra 500ms seems too expensive.
> We should look at a way of doing this in a more 'efficient way' from time
> perspective, may be spinning while the deleteCgroup() cannot be done with a
> minimal sleep and a timeout.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
