[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13739540#comment-13739540 ] Hudson commented on YARN-1036: -- FAILURE: Integrated in Hadoop-Hdfs-0.23-Build #699 (See [https://builds.apache.org/job/Hadoop-Hdfs-0.23-Build/699/]) YARN-1036. Distributed Cache gives inconsistent result if cache files get deleted from task tracker. Contributed by Mayank Bansal and Ravi Prakash (jlowe: http://svn.apache.org/viewcvs.cgi/?root=Apache-SVNview=revrev=1513636) * /hadoop/common/branches/branch-0.23/hadoop-yarn-project/CHANGES.txt * /hadoop/common/branches/branch-0.23/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/main/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/localizer/LocalResourcesTrackerImpl.java * /hadoop/common/branches/branch-0.23/hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-nodemanager/src/test/java/org/apache/hadoop/yarn/server/nodemanager/containermanager/localizer/TestLocalResourcesTrackerImpl.java Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Fix For: 0.23.10 Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13738337#comment-13738337 ] Jason Lowe commented on YARN-1036: -- Agree with Ravi that we should focus on porting the change to 0.23 and fix any issues that also apply to trunk/branch-2 in a separate JIRA. Therefore I agree with Omkar that we should simply break or omit the LOCALIZED case from the switch statement since 0.23 doesn't have localCacheDirectoryManager to match the trunk behavior. Otherwise patch looks good to me. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13738541#comment-13738541 ] Omkar Vinit Joshi commented on YARN-1036: - +1 ... thanks for updating the patch..lgtm. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13738754#comment-13738754 ] Jason Lowe commented on YARN-1036: -- +1 lgtm as well. Committing this. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13733639#comment-13733639 ] Ravi Prakash commented on YARN-1036: Hi Omkar! Thanks a lot for pointing out the problem in the earlier patch. Regarding the changes you are proposing, I meant for this JIRA to simply be a backport of MAPREDUCE-4342. I wasn't able to re-open that JIRA because it has already been closed (hence I had to file this new JIRA). If you have spotted a problem with the current patch, I would welcome your suggested changes. However if you have an issue with the approach, I would request you to please pursue them in a separate JIRA as they lie outside the scope of simple backporting. Most of this code is already in trunk as is. Please let me know if this is acceptable to you. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13731945#comment-13731945 ] Koji Noguchi commented on YARN-1036: When reading MAPREDUCE-4342 initially, I didn't understand why the files were deleted in the first place. Until recently, when we started seeing inconsistent distributed cache after we enabled Disk-Fail-In-Place(MAPREDUCE-2413). NodeManager/TaskTracker keeps on running even after finding bad disks without throwing away distributed cache entries coming from those disks resulting in incomplete set of distributed cache available to our users. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732125#comment-13732125 ] Ravi Prakash commented on YARN-1036: Thanks for the review Omkar and Koji! Are you satisfied with Koji's explanation Omkar? Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732214#comment-13732214 ] Omkar Vinit Joshi commented on YARN-1036: - [~knoguchi] We get LOCALIZED even when resource is localized by public/private localizer (and when resource is in DOWNLOADING state).. so this event will be fired only when file is successfully localized. Probably we need to delete the file only if it is already localized ... thoughts? isResourcePresent is checking for already LOCALIZED file right? Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732637#comment-13732637 ] Ravi Prakash commented on YARN-1036: Omkar! We don't want to delete a file. We want it to be localized again if it is detected to no longer exist. Which is what this patch is doing. This is what Koji is also saying. I already tested the patch does the right thing on my single node cluster. My test methodology was this. 1. Configure 2 directories for the NM. One on the node hard disk, and another on a pen drive. 2. Ran a sleep job with -files option specifying a file. 3. Make sure the file is localized on the pen drive. (If it isn't run another sleep job with a different file to be put in distcache) 4. unplug the pendrive (to simulate a bad disk). Before the patch, running a sleep job requesting the same old file in distcache didn't localize the file again. So if the job had required that file, it would have failed. After the patch, it detects that the file which was already localized is missing, and so it localizes it again. This is the behavior we want. Do you agree? Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732653#comment-13732653 ] Omkar Vinit Joshi commented on YARN-1036: - [~raviprak] Let me try to explain again. * Resource R1 is not in the cache. so on (REQUEST) new resource was created in (DOWNLOADING) state and download started. * Once download is finished; we get (LOCALIZED) (here we don't want to check whether file is present on disc or not.. do we have to? if yes then probably we may have to update trunk also as I separated the logic in one of the jiras.) and we change the resource state to (LOCALIZED) and notify all waiting containers. [File is present on disc]. * Now file got deleted. * Another container requests the file (REQUEST).. now we should definitely check if the file is already present on disc or not. If not remove it from local cache (DataStructure-- by delete the file I wanted to say from local cache not from disc as it is already gone from there). Let me know your thoughts. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732731#comment-13732731 ] Ravi Prakash commented on YARN-1036: [~ojoshi] bq. (here we don't want to check whether file is present on disc or not.. do we have to? I agree that the very first time the file is localized, checking for the existence of the file all over again is redundant. The file could go missing right after checking, and we would be none the wiser. However in 0.23, I am willing to check again. That's why I am not proposing the change to trunk. There's a lot of changes between trunk/branch-2 and branch-0.23 and we are only cherry-picking those necessary for maintaining the line. That is why we are also not backporting all the cool stuff you have done e.g. in YARN-539 and YARN-467 etc. bq. now we should definitely check if the file is already present on disc or not. If not remove it from local cache (DataStructure-- by delete the file I wanted to say from local cache not from disc as it is already gone from there). Isn't that exactly what the new code in the patch is doing? {code} if (rsrc != null (!isResourcePresent(rsrc))) { LOG.info(Resource + rsrc.getLocalPath() + is missing, localizing it again); localrsrc.remove(req); rsrc = null; } {code} Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732738#comment-13732738 ] Ravi Prakash commented on YARN-1036: In the way you suggest {quote} {code} case LOCALIZED: break; case REQUEST: + if (rsrc != null (!isResourcePresent(rsrc))) { +LOG.info(Resource + rsrc.getLocalPath() ++ is missing, localizing it again); +localrsrc.remove(req); +rsrc = null; + } . {code} {quote} Once R1 had been localized, and the disk went bad, we would never again check if it still exists. That's entirely the problem we are trying to solve here. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732743#comment-13732743 ] Omkar Vinit Joshi commented on YARN-1036: - [~raviprak] no it is actually not doing it the way it has to. ResourceState :- DOWNLOADING and LOCALIZED event Do you think there is any point in executing below code? *if (rsrc.getState() == ResourceState.LOCALIZED) {* {code} if (rsrc != null (!isResourcePresent(rsrc))) { LOG.info(Resource + rsrc.getLocalPath() + is missing, localizing it again); localrsrc.remove(req); decrementFileCountForLocalCacheDirectory(req, rsrc); rsrc = null; } if (null == rsrc) { rsrc = new LocalizedResource(req, dispatcher); localrsrc.put(req, rsrc); } public boolean isResourcePresent(LocalizedResource rsrc) { boolean ret = true; if (rsrc.getState() == ResourceState.LOCALIZED) { File file = new File(rsrc.getLocalPath().toUri().getRawPath(). toString()); if (!file.exists()) { ret = false; } } return ret; } {code} anyway lets see what others have to say... but clearly by mixing this we may end up seeing some random race condition issues. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732824#comment-13732824 ] Hadoop QA commented on YARN-1036: - {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12596727/YARN-1036.branch-0.23.patch against trunk revision . {color:red}-1 patch{color}. The patch command could not apply the patch. Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1671//console This message is automatically generated. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13732887#comment-13732887 ] Omkar Vinit Joshi commented on YARN-1036: - Thanks for updating it... few more comments. bq. rsrc == null (I can't fathom why it would be), but still. I don't think it will ever be null in LOCALIZED case. lets isolate them. bq. + } //No break here. Falling through. lets put break; Regarding test code bq. + LocalizedResource lr1 = createLocalizedResource(req1, dispatcher); why are we creating LocalizedResource? Probably we don't need to create one and control explicitly.. We can monitor dispatch queue. thoughts? bq. + file:///tmp/r1), 1); lets not have hard coded paths.. will fail on windows (as you are also creating one file). Also try to create file in current working directory instead. probably you can take a look at TestLocalResourcesTrackerImpl#testHierarchicalLocalCacheDirectories (trunk) Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch, YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13731470#comment-13731470 ] Hadoop QA commented on YARN-1036: - {color:red}-1 overall{color}. Here are the results of testing the latest attachment http://issues.apache.org/jira/secure/attachment/12596459/YARN-1036.branch-0.23.patch against trunk revision . {color:red}-1 patch{color}. The patch command could not apply the patch. Console output: https://builds.apache.org/job/PreCommit-YARN-Build/1664//console This message is automatically generated. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (YARN-1036) Distributed Cache gives inconsistent result if cache files get deleted from task tracker
[ https://issues.apache.org/jira/browse/YARN-1036?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13731473#comment-13731473 ] Omkar Vinit Joshi commented on YARN-1036: - Thanks [~raviprak] Probably we need to isolate the logic for LOCALIZED and REQUEST scenarios? thoughts? {code} + if (rsrc != null (!isResourcePresent(rsrc))) { +LOG.info(Resource + rsrc.getLocalPath() ++ is missing, localizing it again); +localrsrc.remove(req); +rsrc = null; + } {code} the code is not required to be executed when a resource is getting LOCALIZED.. in trunk we have isolated them. Probably as in branch 0.23 we don't have anything like localCacheDirectoryManager it makes sense to just keep break...and do nothing in case it is LOCALIZED? {code} case LOCALIZED: break; case REQUEST: + if (rsrc != null (!isResourcePresent(rsrc))) { +LOG.info(Resource + rsrc.getLocalPath() ++ is missing, localizing it again); +localrsrc.remove(req); +rsrc = null; + } . {code} didn't review the test code. Distributed Cache gives inconsistent result if cache files get deleted from task tracker - Key: YARN-1036 URL: https://issues.apache.org/jira/browse/YARN-1036 Project: Hadoop YARN Issue Type: Bug Components: nodemanager Affects Versions: 0.23.9 Reporter: Ravi Prakash Assignee: Ravi Prakash Attachments: YARN-1036.branch-0.23.patch This is a JIRA to backport MAPREDUCE-4342. I had to open a new JIRA because that one had been closed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira