-
Bug
-
Resolution: Cannot Reproduce
-
Medium
-
None
-
None
-
None
-
None
If Rackspace is doing a “soft” shutdown that invokes the shutdown scripts through init.d, then it could take on the order of 10 minutes to shut down the CDAP master process, as it saves some amount of state before finally shutting down.
What we typically do in the labs and for production redeploys is:
1. ungraceful shutdown of all the CDAP/Hadoop processes: `sudo killall -9 java`
2. wipe the cinder volumes: `sudo rm -rf /opt/data/*`
3. then invoke the undeploy/destroy.
Note that if we are not reusing Cinder volumes (destroying and recreating every time), then you can probably safely skip step 2.
We would like to improve the current behavior in a future release.
- mentioned in
-
Page Loading...