You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafodion.apache.org by "Arvind Narain (JIRA)" <ji...@apache.org> on 2017/04/12 18:16:41 UTC
[jira] [Reopened] (TRAFODION-1989) after uninstall, and reinstall
again, dcscheck report dcs master not up
[ https://issues.apache.org/jira/browse/TRAFODION-1989?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Arvind Narain reopened TRAFODION-1989:
--------------------------------------
Assignee: (was: Amanda Moran)
Reopening to fix this issue in Ambari installer. If the trafodion userid changes then jps command doesn't work resulting in check scripts giving wrong info.
Bash installer has the following change:
function fixPermissions {
#Change ownership of this file to be owned by traf user and traf group
#Errors with JPS will happen if not modified
if [[ -d /tmp/hsperfdata_trafodion ]]; then
$TRAF_PDSH sudo chown -R $TRAF_USER.trafodion /tmp/hsperfdata_trafodion
fi
}
pyinstaller has the following change (including rm -rf during uninstall):
# change permission for hsperfdata
if os.path.exists(TRAF_HSPERFDATA_FILE):
run_cmd('chown -R %s:%s %s' % (traf_user, traf_group, TRAF_HSPERFDATA_FILE))
===
Problem seen in Ambari env (dcs master is up but shows as down):
[trafodion@rmarton-2 tmp]$ ls -ld hsper*
drwxr-xr-x 2 ambari-qa hadoop 4096 Apr 10 23:36 hsperfdata_ambari-qa
drwxr-xr-x 2 hbase hadoop 4096 Apr 10 14:56 hsperfdata_hbase
drwxr-xr-x 2 hcat hadoop 4096 Apr 10 14:45 hsperfdata_hcat
drwxr-xr-x 2 hdfs hadoop 4096 Apr 10 14:38 hsperfdata_hdfs
drwxr-xr-x 2 hive hadoop 4096 Apr 10 14:45 hsperfdata_hive
drwxr-xr-x 2 mapred hadoop 4096 Apr 10 14:44 hsperfdata_mapred
drwxr-xr-x 2 root root 4096 Apr 10 21:41 hsperfdata_root
drwxr-xr-x 2 1003 trafodion 4096 Apr 10 20:21 hsperfdata_trafodion
drwxr-xr-x 2 yarn hadoop 4096 Apr 10 14:43 hsperfdata_yarn
drwxr-xr-x 2 zookeeper hadoop 4096 Apr 10 14:49 hsperfdata_zookeeper
[trafodion@rmarton-2 tmp]$ id
uid=512(trafodion) gid=503(trafodion) groups=503(trafodion)
[trafodion@rmarton-2 tmp]$ jps
[trafodion@rmarton-2 tmp]$ id
uid=512(trafodion) gid=503(trafodion) groups=503(trafodion)
[trafodion@rmarton-2 tmp]$ dcscheck
DcsMaster is not started. Please start DCS using 'dcsstart' command...
Process Configured Actual Down
--------- ---------- ------ ----
DcsMaster 2 0 2
DcsServer 2 0 2
mxosrvr 16 16
[trafodion@rmarton-2 tmp]$ cdw
[trafodion@rmarton-2 apache-trafodion_server-2.1.0]$ cd dcs-2.1.0/
[trafodion@rmarton-2 dcs-2.1.0]$ bin/zkcli
-bash: bin/zkcli: No such file or directory
[trafodion@rmarton-2 dcs-2.1.0]$ bin/dcs zkcli
Connecting to rmarton-2.novalocal:2181
Welcome to ZooKeeper!
JLine support is enabled
WATCHER::
WatchedEvent state:SyncConnected type:None path:null
[zk: rmarton-2.novalocal:2181(CONNECTED) 0] ls /trafodion/dcs/master
[rmarton-2.novalocal:23400:100:1491861269695]
[zk: rmarton-2.novalocal:2181(CONNECTED) 1] quit
Quitting...
[trafodion@rmarton-2 dcs-2.1.0]$
> after uninstall, and reinstall again, dcscheck report dcs master not up
> -----------------------------------------------------------------------
>
> Key: TRAFODION-1989
> URL: https://issues.apache.org/jira/browse/TRAFODION-1989
> Project: Apache Trafodion
> Issue Type: Bug
> Components: installer
> Reporter: liu ming
>
> in the same cluster, one install a trafodion version.
> then use trafodion_uninstaller to remove the installation.
> later at same cluster, install again via trafodion_install
> after the new installation, dcscheck fail.
> jps cannot show related DCS java processes, but those processes are up and running.
> It is due to a legacy /tmp/hsperfdata_trafodion direcotry belong to the previous trafodion user. The new user cannot write into this folder, but jps rely on it to work correct. And dcscheck rely on jps to work correctly.
> Either installer remove that legacy directory, or uninstaller remove it.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)