Marcello,
Not sure if this is recommended, but this is what I'm using to start/stop
MCF on CentOS 6. Seems to work fairly well, but I'm not in a production
environment at this point.
I know it doesn't directly address the issue with the agent JVM pid being
killed outside of stop-agents.sh, but adding nohup into the mix minimizes
the likelihood that that happens, to me, anyway.
start.sh:
#!/bin/bash -e
nohup ./start-agents.sh > agents.out 2>&1&
nohup ./start-webapps.sh > webapps.out 2>&1&
echo $! > webapps.pid
stop.sh:
#!/bin/bash -e
./stop-agents.sh
pid=$(<"webapps.pid")
if [[ -z "$pid" ]] ; then
echo No webapps pid.
exit 1
fi
gid=$(ps opgid= "$pid")
pkill -TERM -g $gid
Mark
On Wed, Nov 6, 2013 at 8:09 AM, Marcello Lorenzi <[email protected]> wrote:
> Hi All,
> I'm trying to install and testing the Manifold agent on a clustered
> environment based on CentOS 6, and I noticed a strange behavior of agent
> when the JVM PID were killed manually without using the stop-agents.sh
> scripts. In fact after the kill if I try to start the agent I noticed
> this messages "Agent already in use" and I couldn't start the agent.
>
> The only solution is to run locks-clean.sh script to mitigate the issue,
> but to run this script I have to stop the agent and the webconsole.
>
> Is it a correct procedure to cleanup the environment ?
>
> Thanks,
> Marcello
>
>
>
>
