Last modified: 2014-02-03 03:00:45 UTC
My script: http://pastebin.com/zGDtjC64 Based on output and file existence, it seems every failed mysqldump call (due to missing database) takes one minute to end, while it's less than one second on tools-dev.
I removed my previous job and started a new one, and tools-exec-02 is allocated. Now the script runs smoothly. The previous host was tools-exec-01. btw In meantime, I changed the script a bit, removing " 2> /dev/null" in order to observe whether it's some abnormal error. Hope that it doesn't matter much.
Is this still an issue?
(In reply to comment #2) > Is this still an issue? I haven't observed such events recently, based on file timestamps, but I can't know whether tools-exec-01 has been allocated for them.
(In reply to comment #3) > [...] > I haven't observed such events recently, based on file timestamps, but I > can't > know whether tools-exec-01 has been allocated for them. You can force executing a job on a specific host by requesting the "hostname" resource, i. e. "jsub -l hostname=tools-exec-03 true" will run "true" on tools-exec-03.
(In reply to comment #2) > Is this still an issue? ok it seems better now.
