dcsstop does not stop all mxosrvrs

[Aruna Sadashiva]

All MXOSRVRs do not get stopped when dcsstop (or sqstop or even ckillall) is run.

cstat still shows several processes like this:
rhel-mapr3: trafodion 9255 7827 futex_ 1537324 3146568 01:30:06 Sl mxosrvr -ZkHost rhel-mapr1:2181,rhel-mapr2:2181,rhel-mapr3:2181 -RZ rhel-mapr3.hpl.hp.com:3:3 -ZkPnode /trafodion -CNGTO 60 -ZKSTO 180 -EADSCO 0 -TCPADD 15.25.112.36 -MAXHEAPPCT 0

Workaround from dev is to run this :

pdsh $MY_NODES "ps -u $USER -o pid,cmd | grep mxosrvr " |sed -e "s@:@@" | awk '{print "ssh "$1" kill -9 "$2}'>killmxo
killmxo

[Trafodion-Gerrit (neo-devtools)]

Fix proposed to branch: master
Review: https://review.trafodion.org/967

[Trafodion-Gerrit (neo-devtools)]

Reviewed: https://review.trafodion.org/967
Committed: https://github.com/trafodion/core/commit/ddf46a489733fa52a8a5a3a3250fb7d48525f691
Submitter: Trafodion Jenkins
Branch: master

commit ddf46a489733fa52a8a5a3a3250fb7d48525f691
Author: Tharaknath Capirala <email address hidden>
Date: Thu Jan 15 08:24:45 2015 +0000

    Fix for bug 1404108 where mxosrvrs do not exit on a sqstop

    When sqstop or dcsstop is called some mxosrvrs do not exit when the
    statistics collection feature is enabled. The statistics collection
    occurs in a separate thread and that thread sometimes does not
    terminate and also holds on to a mutex and does not get released during
    exit. The fix is to retry the thread termination during exit time.

    Fixes bug 1404108

    Change-Id: Ia5764a4e3672da8dbade00f8e4b5bdacbb0d5ac5

[Aruna Sadashiva (aruna-sadashiva)]

dcsstop works ok in recent builds.