Fedora Account System
Red Hat Associate
Red Hat Customer
Description of problem: After the ambari wizard installation the /mnt/brick1/hadoop/yarn/ directory is owned by root:root and thus the timeline service cannot start. The owner needs to be yarn:hadoop. Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
enable_vol.sh fixes the perms and owner of /mnt/brick1/hadoop/yarn and all files under via chown -R and chmod -R.
When/how/by what should be created directory /mnt/brick1/hadoop/yarn on Master server? If I'll configure yarn.timeline-service.leveldb-timeline-store.path to point to /mnt/brick1/hadoop/yarn/timeline (as it is in our actual documentation), enabling volume fails with following error: DEBUG: update yarn local directories on MASTER (yarn-master)... ERROR: chown/chmod local yarn dir "/mnt/brick1/hadoop/yarn": /mnt/brick1/hadoop/yarn missing on MASTER I think this is related to the bug 1160608. Our automatic tests were using "/var/log/hadoop-yarn/timeline" as it is suggested in Hortonworks documentation (bug 1160608 comment 0). But now I've check the installation process and correct it accordingly to our documentation and now everything is failing. So moving this BZ back to ASSIGNED, to evaluate, where should yarn.timeline-service.leveldb-timeline-store.path point. And if /mnt/brick1/hadoop/yarn/timeline is correct location, than evaluate who should create /mnt/brick1/hadoop/yarn directory (or set proper permission on the parent directory). If it points to /var/log/hadoop-yarn/timeline, the directory /var/log/hadoop-yarn/ is owned by yarn:hadoop (from installation of hadoop-yarn package), so yarn is able to create required timeline files/directories itself.
Re comment #5: I am unable to duplicate this failure. Can you please paste the /var/log/rhs-hadoop-install.log file showing the entire enable_vol.sh portion? Note: per my comments #11 in bz 1160608, Ambari creates this directory.
Jeff, you are right, sorry for false alarm. It was probably some misconfiguration on my site. When I've rerun the tests and check the correct configuration accordingly to Doc, it works properly. >> MODIFIED
Tested and VERIFIED on Red Hat Storage Server 3.0 Update 4 with rhs-hadoop-install-2_44-1.el6rhs.noarch. Ambari correctly created directories /mnt/brick1/hadoop/yarn and enable_vol.sh change the ownership and permissions: # ll -d /mnt/brick1/hadoop/yarn/ drwxr-xr-x. 5 yarn hadoop 4096 Mar 17 09:57 /mnt/brick1/hadoop/yarn/ # ll -d /mnt/brick1/hadoop/yarn/timeline/ drwxr-xr-x. 3 yarn hadoop 4096 Mar 17 09:57 /mnt/brick1/hadoop/yarn/timeline/ >> VERIFIED
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHEA-2015-0761.html