We've been seeing intermittent failures when
stopping a member where it times out waiting
for the karaf process to disappear. Eg
https://logs.opendaylight.org/releng/vex-yul-odl-jenkins-1/controller-csit-3node-clustering-tell-all-neon/107/robot-plugin/log.html.gz#s1-s17
In that case, it timed out at 08:45:10 - the
karaf log indicates it stopped somehwere between
08:45:11 and 8:45:32 when it was restarted. So
the 2 min deadline wasn't quite enough in this
case.
I think at least part of the problem is that
CSIT installs the "world" or at least all the
managed projects which adds to the startup and
shutdown time - coupled with random delays due to
workload in the CSIT env. In lieu of trimming down
the installed features, let's double the timeout
to 4 min to (hopefully) give plenty of cushion.
Change-Id: I1d3b85ec1600c26c6ee187ffc5036a513151f2c6
Signed-off-by: Tom Pantelis <tompantelis@gmail.com>
[Return] ${updated_index_list}
Stop_Members_From_List_Or_All
- [Arguments] ${member_index_list}=${EMPTY} ${original_index_list}=${EMPTY} ${confirm}=True ${timeout}=120s
+ [Arguments] ${member_index_list}=${EMPTY} ${original_index_list}=${EMPTY} ${confirm}=True ${timeout}=240s
[Documentation] If the list is empty, stops all ODL instances. Otherwise stop members based on \${stop_index_list}
... If \${confirm} is True, verify stopped instances are not there anymore.
... The KW will return a list of available members: \${updated index_list}=\${original_index_list}-\${member_index_list}