Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
0.4.0
-
None
-
None
-
redhat enterprise hup signal agent collector
Description
On RHEL 5.4, agents and collectors do not respond to HUP signals sent to the process PID.
I start the agent like this, then attempt to kill it like so:
$ pwd /usr/local/chukwa/bin $ ./start-agents.sh $ pgrep -lf Agent 23148 /usr/java/latest/bin/java -Djava.library.path=/usr/lib/hadoop/lib/native/Linux-X86-64 [...] org.apache.hadoop.chukwa.datacollection.agent.ChukwaAgent $ cat /var/run/Agent.pid 23148 $ kill -hup 23148 $ ps -fp 23148 UID PID PPID C STIME TTY TIME CMD root 23148 1 0 Jan23 pts/0 00:00:19 /usr/java/latest/bin/java -Djava.library.path=/usr/lib/hadoop/lib/native/}} $ pgrep -lf Agent 23148 /usr/java/latest/bin/java -Djava.library.path=/usr/lib/hadoop/lib/native/Linux-X86-64 [...] : org.apache.hadoop.chukwa.datacollection.agent.ChukwaAgent}}
The agent is still running, and /var/log/agent.log shows no indication that the signal was seen.
The agent is running one adaptor:
$ echo "list" | nc localhost 9093 adaptor_b5382452cfdc2ae81bd45d8126804843) org.apache.hadoop.chukwa.datacollection.adaptor.ExecAdaptor Iostat 60 /usr/bin/iostat -x -k 55 2 11215696
We can see this child process running too:
$ ps -f --ppid 23148 UID PID PPID C STIME TTY TIME CMD root 7116 23148 0 12:40 pts/0 00:00:00 /usr/bin/iostat -x -k 55 2
In the case of the collector, I start it like so:
$ ./start-collectors.sh localhost: starting collector, logging to /var/log/chukwa-chukwa-collector-<hostname.>out localhost: 2012-01-24 04:24:11.04::INFO: Logging to STDERR via org.mortbay.log.StdErrLog localhost: 2012-01-24 04:24:11.104::INFO: jetty-6.1.11 $ pgrep -lf collector 10220 /usr/java/latest/bin/java -Djava.library.path=/usr/lib/hadoop/lib/native/Linux-X86-64 -DCHUKWA_HOME=/usr/local/chukwa/bin/.. [...] : org.apache.hadoop.chukwa.datacollection.collector.CollectorStub $ kill -HUP 10220 $ pgrep -lf collector 10220 /usr/java/latest/bin/java -Djava.library.path=/usr/lib/hadoop/lib/native/Linux-X86-64 -DCHUKWA_HOME=/usr/local/chukwa/bin/ [...]: org.apache.hadoop.chukwa.datacollection.collector.CollectorStub
The logs on the collector show no indication that any signal was received.