[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184468#comment-14184468 ] Hudson commented on YARN-2183: -- SUCCESS: Integrated in Hadoop-Yarn-trunk #724 (See [https://builds.apache.org/job/Hadoop-Yarn-trunk/724/]) YARN-2183. [YARN-1492] Cleaner service for cache manager. (Chris Trezzo and Sangjin Lee via kasha) (kasha: rev c51e53d7aad46059f52d4046a5fedfdfd3c37955) * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/TestCleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/TestInMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerService.java * hadoop-yarn-project/CHANGES.txt * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/SharedCacheManager.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/InMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/SCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/CleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/TestCleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common/src/main/java/org/apache/hadoop/yarn/server/sharedcache/SharedCacheUtil.java Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Fix For: 2.7.0 Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184499#comment-14184499 ] Hudson commented on YARN-2183: -- SUCCESS: Integrated in Hadoop-Hdfs-trunk #1913 (See [https://builds.apache.org/job/Hadoop-Hdfs-trunk/1913/]) YARN-2183. [YARN-1492] Cleaner service for cache manager. (Chris Trezzo and Sangjin Lee via kasha) (kasha: rev c51e53d7aad46059f52d4046a5fedfdfd3c37955) * hadoop-yarn-project/CHANGES.txt * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerService.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/SCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/SharedCacheManager.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/CleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/TestCleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common/src/main/java/org/apache/hadoop/yarn/server/sharedcache/SharedCacheUtil.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/TestCleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/InMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/TestInMemorySCMStore.java Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Fix For: 2.7.0 Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184512#comment-14184512 ] Hudson commented on YARN-2183: -- FAILURE: Integrated in Hadoop-Mapreduce-trunk #1938 (See [https://builds.apache.org/job/Hadoop-Mapreduce-trunk/1938/]) YARN-2183. [YARN-1492] Cleaner service for cache manager. (Chris Trezzo and Sangjin Lee via kasha) (kasha: rev c51e53d7aad46059f52d4046a5fedfdfd3c37955) * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/TestInMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/TestCleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common/src/main/java/org/apache/hadoop/yarn/server/sharedcache/SharedCacheUtil.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/SCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/SharedCacheManager.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/TestCleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerService.java * hadoop-yarn-project/CHANGES.txt * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/CleanerMetrics.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/InMemorySCMStore.java Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Fix For: 2.7.0 Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184198#comment-14184198
]
Hadoop QA commented on YARN-2183:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12677118/YARN-2183-trunk-v8.patch
against trunk revision f44cf99.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 3 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}. There were no new javadoc 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 2.0.3) 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-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5563//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5563//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch,
YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch,
on-demand-cleaner.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184213#comment-14184213 ] Karthik Kambatla commented on YARN-2183: +1. Checking this in.. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184216#comment-14184216 ] Hudson commented on YARN-2183: -- FAILURE: Integrated in Hadoop-trunk-Commit #6344 (See [https://builds.apache.org/job/Hadoop-trunk-Commit/6344/]) YARN-2183. [YARN-1492] Cleaner service for cache manager. (Chris Trezzo and Sangjin Lee via kasha) (kasha: rev c51e53d7aad46059f52d4046a5fedfdfd3c37955) * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common/src/main/java/org/apache/hadoop/yarn/server/sharedcache/SharedCacheUtil.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-common/src/main/resources/yarn-default.xml * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/TestInMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerService.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/CleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/TestCleanerTask.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-api/src/main/java/org/apache/hadoop/yarn/conf/YarnConfiguration.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/InMemorySCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/SharedCacheManager.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/store/SCMStore.java * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/main/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/CleanerMetrics.java * hadoop-yarn-project/CHANGES.txt * hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/src/test/java/org/apache/hadoop/yarn/server/sharedcachemanager/metrics/TestCleanerMetrics.java Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Fix For: 2.7.0 Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14184369#comment-14184369 ] Sangjin Lee commented on YARN-2183: --- Thanks Karthik! Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Fix For: 2.7.0 Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, YARN-2183-trunk-v8.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183486#comment-14183486 ] Sangjin Lee commented on YARN-2183: --- Yes, I would arrive at essentially the same result if we removed this race condition check. This would ensure that only one cleaner task runs at a given time (i.e. serial execution). What it does not prevent, however, is a *back-to-back* execution if an on-demand cleaner run was submitted close to a scheduled cleaner run. The key here is that we're using a *single-threaded* scheduled executor. Here is the sequence with your patch: # a scheduled run gets under way and sets cleanerTaskIsRunning to true # an on-demand run gets submitted: but it will not run until the scheduled run finished because it runs on a single-threaded scheduled executor # the scheduled run finishes and flips cleanerTaskIsRunning to false # the on-demand run gets under way, and finds cleanerTaskIsRunning is false so it runs again (i.e. a back-to-back run) With v.6: # a scheduled run gets under way and sets cleanerTaskIsRunning to true # an on-demand run is requested but does *not* get scheduled because we're checking if a run is already in progress before submitting it (i.e. no back-to-back run) # the scheduled run finishes and flips cleanerTaskIsRunning to false Another way of accomplishing this is to put the on-demand cleaner task on a different executor or a different thread within the same scheduled executor (e.g. giving the scheduled executor 2 threads instead of 1). Then we could do this with the simpler code as the check can be made concurrently and it will do the right thing. What do you think? Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183571#comment-14183571 ] Karthik Kambatla commented on YARN-2183: Ah, now I get it. I missed the executor serializing the tasks already. I think it is okay to use 2 threads in the executor. Otherwise, we are having to differentiate between scheduled and impromptu cleaner tasks, which is probably an overkill to achieve this. We should probably add a comment to explain the choice of 2 threads, which is to reduce the chances of running cleaner-tasks back-to-back. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183574#comment-14183574 ] Sangjin Lee commented on YARN-2183: --- OK, let me update the patch. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch, on-demand-cleaner.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183800#comment-14183800
]
Hadoop QA commented on YARN-2183:
-
{color:red}-1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12677057/YARN-2183-trunk-v7.patch
against trunk revision 683897f.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 3 new
or modified test files.
{color:red}-1 javac{color}. The applied patch generated 1294 javac
compiler warnings (more than the trunk's current 1271 warnings).
{color:green}+1 javadoc{color}. There were no new javadoc 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 2.0.3) warnings.
{color:green}+1 release audit{color}. The applied patch does not increase
the total number of release audit warnings.
{color:red}-1 core tests{color}. The test build failed in
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5554//testReport/
Javac warnings:
https://builds.apache.org/job/PreCommit-YARN-Build/5554//artifact/patchprocess/diffJavacWarnings.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5554//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch,
YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, on-demand-cleaner.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183826#comment-14183826
]
Sangjin Lee commented on YARN-2183:
---
Something funny about the latest jenkins run. When I run the javac warning test
myself, there are 0 javac warnings from the sharedcache-manager project:
{noformat}
[INFO]
[INFO] Building hadoop-yarn-server-sharedcachemanager 3.0.0-SNAPSHOT
[INFO]
[INFO]
[INFO] --- maven-clean-plugin:2.4.1:clean (default-clean) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Deleting
/Users/sjlee/git/hadoop-woodpecker/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/target
[INFO]
[INFO] --- maven-antrun-plugin:1.7:run (create-testdirs) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Executing tasks
main:
[mkdir] Created dir:
/Users/sjlee/git/hadoop-woodpecker/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/target/test-dir
[INFO] Executed tasks
[INFO]
[INFO] --- maven-resources-plugin:2.2:resources (default-resources) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Using default encoding to copy filtered resources.
[INFO]
[INFO] --- maven-compiler-plugin:2.5.1:compile (default-compile) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Compiling 10 source files to
/Users/sjlee/git/hadoop-woodpecker/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/target/classes
[INFO]
[INFO] --- maven-resources-plugin:2.2:testResources (default-testResources) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Using default encoding to copy filtered resources.
[INFO]
[INFO] --- maven-compiler-plugin:2.5.1:testCompile (default-testCompile) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Compiling 4 source files to
/Users/sjlee/git/hadoop-woodpecker/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/target/test-classes
[INFO]
[INFO] --- maven-jar-plugin:2.3.1:test-jar (default) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Building jar:
/Users/sjlee/git/hadoop-woodpecker/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager/target/hadoop-yarn-server-sharedcachemanager-3.0.0-SNAPSHOT-tests.jar
[INFO]
[INFO] --- maven-surefire-plugin:2.16:test (default-test) @
hadoop-yarn-server-sharedcachemanager ---
[INFO] Tests are skipped.
[INFO]
{noformat}
I'm going to rekick jenkins...
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch,
YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, on-demand-cleaner.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14183849#comment-14183849
]
Hadoop QA commented on YARN-2183:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12677057/YARN-2183-trunk-v7.patch
against trunk revision 683897f.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 3 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}. There were no new javadoc 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 2.0.3) 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-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5556//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5556//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch,
YARN-2183-trunk-v6.patch, YARN-2183-trunk-v7.patch, on-demand-cleaner.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14181502#comment-14181502
]
Sangjin Lee commented on YARN-2183:
---
Thanks for the review [~kasha]!
{quote}
I understand we need a check to prevent the race. I wonder if we can just
re-use the existing check in CleanerTask#run instead of an explicit check in
CleanerService#runCleanerTask? From what I remember, that would make the code
in CleanerTask#run cleaner as well. (no pun)
{quote}
The main motivation for this somewhat elaborate double check was for a
situation when an on-demand cleaner run comes in and a scheduled cleaner run
starts. Without this check, we would have two cleaner runs back to back which
is somewhat wasteful.
Having said that, I think it is debatable how important it is to avoid that
situation and whether it is an optimization worth doing. One could argue that
this is bit too fine optimization. Thoughts?
{quote}
I poked around a little more, and here is what I think. SharedCacheManager
creates an instance of AppChecker, rest of the SCM pieces (Store,
CleanerService) should just use the same instance. This instance can be passed
either in the constructor or through an SCMContext similar to RMContext. Or, we
could add SCM#getAppChecker.
In its current form, CleanerTask#cleanResourceReferences fetches the references
from the store, checks if the apps are running, and asks the store to remove
the references. Moving the whole method to the store would simplify the code
more.
{quote}
Yes, I agree that moving cleanResourceReferences() to the store would simplify
code here. There is one caveat however. Currently
CleanerTask.cleanResourceReferences() is generic: i.e. it does not depend on
the type of the store. But if we move this to the store, then I think it would
need to be abstract at the level of SCMStore and each store implementation
would need to implement its own. The main reason is that the concurrency/safety
semantics would be different from store impl to store impl. In the case of the
in-memory store, it would use the synchronization on the interned key. But in
case of other stores, that does not apply and they need to do their own
implementation mostly because how they handle concurrency will be different. So
it would mean largely copying and pasting of the same logic with a small
difference of how they handle concurrency. That does seem to be a downside of
this approach. What do you think?
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14181684#comment-14181684 ] Karthik Kambatla commented on YARN-2183: bq. Without this check, we would have two cleaner runs back to back which is somewhat wasteful. I don't entirely remember my train of thought, but I can take a look again and see if we can implement it in a simpler way and get the same guarantee. May be, after the next patch. bq. Currently CleanerTask.cleanResourceReferences() is generic: i.e. it does not depend on the type of the store. Can we keep this method as is, but mark it protected. The store implementations can choose to use it. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14181723#comment-14181723 ] Sangjin Lee commented on YARN-2183: --- You mean, moving the method to SCMStore but mark it protected? If so, for CleanerTask to be able to call it, it cannot be protected, right? One thing we can do is to move it to SCMStore as a public method, but let implementations override/augment it. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14181772#comment-14181772 ] Karthik Kambatla commented on YARN-2183: Yes. Sorry for the confusion. Just looked at the code again. My suggestion is to move cleanResourceReferences to SCMStore and mark it @Private public final. Does that make sense? Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14182431#comment-14182431
]
Hadoop QA commented on YARN-2183:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12676769/YARN-2183-trunk-v6.patch
against trunk revision 57dec28.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 3 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}. There were no new javadoc 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 2.0.3) 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-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5536//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5536//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch,
YARN-2183-trunk-v6.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14182443#comment-14182443 ] Sangjin Lee commented on YARN-2183: --- Better link: https://github.com/ctrezzo/hadoop/compare/ctrezzo:trunk...sharedcache-3-YARN-2183-cleaner Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch, YARN-2183-trunk-v6.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14180676#comment-14180676
]
Karthik Kambatla commented on YARN-2183:
bq. we do have a YARN admin command implemented that lets you run the cleaner
task on demand (YARN-2189).
Cool. In that case, I see the merit to keeping runCleanerTask around.
bq. this check is needed to prevent a race (i.e. not allow an on-demand run
when a scheduled run is in progress).
I understand we need a check to prevent the race. I wonder if we can just
re-use the existing check in CleanerTask#run instead of an explicit check in
CleanerService#runCleanerTask? From what I remember, that would make the code
in CleanerTask#run cleaner as well. (no pun)
bq. However, it’s not clear to me whether a dependency from an SCMStore to an
AppChecker is always a fine requirement for other types of stores.
I poked around a little more, and here is what I think. SharedCacheManager
creates an instance of AppChecker, rest of the SCM pieces (Store,
CleanerService) should just use the same instance. This instance can be passed
either in the constructor or through an SCMContext similar to RMContext. Or, we
could add SCM#getAppChecker.
In its current form, CleanerTask#cleanResourceReferences fetches the references
from the store, checks if the apps are running, and asks the store to remove
the references. Moving the whole method to the store would simplify the code
more.
The latest patch looks pretty good but for the above two points. One other nit:
One of {CleanerTask, CleanerService} has unused imports.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14172007#comment-14172007
]
Sangjin Lee commented on YARN-2183:
---
Thanks Karthik for the review. We have addressed most of your review comments
in the updated patch. Items that need more discussion are below.
(CleanerService)
{quote}
runCleanerTask: Instead of checking if there is a scheduled-cleaner-task
running here, why not just rely on the check in CleanerTask#run(). Agree, we
might be doing a little more work here unnecessarily, but not sure the savings
are worth an extra check and an extra parameter in the CleanerTask constructor.
{quote}
We do have the on-demand cleaner feature (YARN-2189; see below for more), and
this check is needed to prevent a race (i.e. not allow an on-demand run when a
scheduled run is in progress).
{quote}
How does a user use runCleanerTask? Instantiate another SCM? The SCM isn't
listening to any requests. I can see the SCM being run in the RM, and one could
potentially add yarn rmadmin -clean-shared-cache. In any case, given there is
no way to reach a running SCM, I would remove runCleanerTask altogether for
now, and add it back later when we need it? Thoughts?
{quote}
As we discussed offline, we do have a YARN admin command implemented that lets
you run the cleaner task on demand (YARN-2189). The admin command implements a
proper ACL check (based on a YARN admin credential), and sends an RPC request
to the running SCM. Since patches are organized this way, it may not have been
very obvious by looking at this patch alone.
{quote}
Should we worry about users starting SCMs with roots at different levels that
can lead to multiple cleaners?
{quote}
In theory it is possible. However, in reality checking that might be being bit
too cautious. I think it might be fine without this check. Let me know what you
think.
(CleanerTask)
{quote}
Should cleanResourceReferences be moved to SCMStore?
{quote}
That’s an interesting suggestion. In fact, with the InMemorySCMStore (which
already has a reference to an AppChecker to clean up the initial apps) it may
be OK to create SCMStore.cleanResourceReferences(). However, it’s not clear to
me whether a dependency from an SCMStore to an AppChecker is always a fine
requirement for other types of stores. In that sense, I would be hesitant to
create this coupling by introducing SCMStore.cleanResourceReferences(). What do
you think?
{quote}
For the race condition (YARN-2663), would it help to handle the delete files on
HDFS in the store#remove?
{quote}
That’s a possibility. However, I think there can be an easier way to fix this
race condition. This is partially due to the way the cleaner task is deleting
unused files. Currently it deletes the entire directory as opposed to the
specific file. The race can be fixed by avoiding deleting the directory. We’ll
add the proper fix later on YARN-2663.
(CleanerMetrics)
{quote}
Make initSingleton private and call it in getInstance if the instance is null?
{quote}
We looked into that, but the difficulty is initSingleton() needs the
configuration which getInstance() does not provide.
{quote}
How about using MutableRate or MutableStat for the rates?
{quote}
We have removed the rate-specific metrics altogether as they can be derived
from the original metrics.
{quote}
Do we need CleanerMetricsCollector, wouldn't CleanerMetrics extending
MetricsSource suffice?
{quote}
For making the metrics available in JMX, indeed CleanerMetrics would suffice.
CleanerMetricsCollector was introduced to create a web UI (HTML) to select a
handful of metrics to show in HTML. That was to come in YARN-2203. Having said
that, I think we can remove CleanerMetricsCollector for now, and not introduce
the web UI. That UI is minimal anyway, and we can consider introducing a more
refined web UI at a later point once this version is released.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14172038#comment-14172038
]
Hadoop QA commented on YARN-2183:
-
{color:green}+1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12674922/YARN-2183-trunk-v5.patch
against trunk revision 1220bb7.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 2 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}. There were no new javadoc 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 2.0.3) 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-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5400//testReport/
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5400//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch, YARN-2183-trunk-v5.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14167093#comment-14167093
]
Hadoop QA commented on YARN-2183:
-
{color:red}-1 overall{color}. Here are the results of testing the latest
attachment
http://issues.apache.org/jira/secure/attachment/12673746/YARN-2183-trunk-v4.patch
against trunk revision cb81bac.
{color:green}+1 @author{color}. The patch does not contain any @author
tags.
{color:green}+1 tests included{color}. The patch appears to include 2 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}. There were no new javadoc 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 2.0.3) warnings.
{color:red}-1 release audit{color}. The applied patch generated 1
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-common
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-sharedcachemanager.
{color:green}+1 contrib tests{color}. The patch passed contrib unit tests.
Test results:
https://builds.apache.org/job/PreCommit-YARN-Build/5363//testReport/
Release audit warnings:
https://builds.apache.org/job/PreCommit-YARN-Build/5363//artifact/patchprocess/patchReleaseAuditProblems.txt
Console output: https://builds.apache.org/job/PreCommit-YARN-Build/5363//console
This message is automatically generated.
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14167105#comment-14167105 ] Sangjin Lee commented on YARN-2183: --- The release audit warning is unrelated. It is about hadoop-yarn-project/hadoop-yarn/hadoop-yarn-registry/src/main/resources/.keep, which is a file that already exists on trunk, and not added by this patch. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[
https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14167175#comment-14167175
]
Karthik Kambatla commented on YARN-2183:
Thanks for updating the patch, Chris.
Review comments on the current patch:
# YarnConfiguration - let us include units for time in the config names;
period-mins, initial-delay-mins, resource-sleep-ms.
# CleanerService
## Don't need a SHUTDOWN_HOOK_PRIORITY specific to CleanerService, it is not
being used anywhere either.
## Nit: In the constructor, can we explicitly initialize AtomicBoolean(false).
## Nit: Rename scheduler to executor or scheduledExecutor?
## serviceStart: the exception thrown should have a message more amenable to
the user. How about appending It appears there is another CleanerService
running in the cluster?
## serviceStop: the log message that the background thread stopped should be
moved to the else-block in the try, instead of after catch?
{code}
try {
if (!scheduler.awaitTermination(10, TimeUnit.SECONDS)) {
LOG.warn(Gave up waiting for the cleaner task to shutdown.);
}
} catch (InterruptedException e) {
LOG.warn(The cleaner service was interrupted while shutting down the
task.,
e);
}
LOG.info(The background thread stopped.);
{code}
## runCleanerTask: Instead of checking if there is a scheduled-cleaner-task
running here, why not just rely on the check in CleanerTask#run(). Agree, we
might be doing a little more work here unnecessarily, but not sure the savings
are worth an extra check and an extra parameter in the CleanerTask constructor.
## How does a user use runCleanerTask? Instantiate another SCM? The SCM isn't
listening to any requests. I can see the SCM being run in the RM, and one could
potentially add yarn rmadmin -clean-shared-cache. In any case, given there is
no way to reach a running SCM, I would remove runCleanerTask altogether for
now, and add it back later when we need it? Thoughts?
## Should we worry about users starting SCMs with roots at different levels
that can lead to multiple cleaners?
#CleanerTask:
## Nit: RENAMED_SUFFIX can be private
## Add @Override to run()
## The following code is duplicated in InMemoryStateStore a well. May be, we
should just add a static method to SharedCacheUtil?
{code}
StringBuilder pattern = new StringBuilder();
for (int i = 0; i nestedLevel; i++) {
pattern.append(*/);
}
pattern.append(*);
{code}
## process: the sleep between directories is in millis, do we want to really
calculate nanos?
## Should cleanResourceReferences be moved to SCMStore?
## For the race condition (YARN-2663), would it help to handle the delete files
on HDFS in the store#remove?
# CleanerMetrics:
## Make initSingleton private and call it in getInstance if the instance is
null?
## How about using MutableRate or MutableStat for the rates?
# Do we need CleanerMetricsCollector, wouldn't CleanerMetrics extending
MetricsSource suffice?
Cleaner service for cache manager
-
Key: YARN-2183
URL: https://issues.apache.org/jira/browse/YARN-2183
Project: Hadoop YARN
Issue Type: Sub-task
Reporter: Chris Trezzo
Assignee: Chris Trezzo
Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch,
YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch
Implement the cleaner service for the cache manager along with metrics for
the service. This service is responsible for cleaning up old resource
references in the manager and removing stale entries from the cache.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14167496#comment-14167496 ] Sangjin Lee commented on YARN-2183: --- Thanks much for great feedback Karthik! I'm going over them, and I'll reply with a new patch and comments. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Commented] (YARN-2183) Cleaner service for cache manager
[ https://issues.apache.org/jira/browse/YARN-2183?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=14164357#comment-14164357 ] Chris Trezzo commented on YARN-2183: Err. I meant attached v4. Cleaner service for cache manager - Key: YARN-2183 URL: https://issues.apache.org/jira/browse/YARN-2183 Project: Hadoop YARN Issue Type: Sub-task Reporter: Chris Trezzo Assignee: Chris Trezzo Attachments: YARN-2183-trunk-v1.patch, YARN-2183-trunk-v2.patch, YARN-2183-trunk-v3.patch, YARN-2183-trunk-v4.patch Implement the cleaner service for the cache manager along with metrics for the service. This service is responsible for cleaning up old resource references in the manager and removing stale entries from the cache. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
