Author Topic: supervisor_flags remove_logs  (Read 2474 times)

Achilles

  • Sr. Member
  • ****
  • Posts: 25
supervisor_flags remove_logs
« on: May 03, 2010, 03:29:46 PM »
hi

the admin manual says that "remove_logs" is the included as default in supervisor_flags.

but if I do a "qbremove zzz" the logs are still there on the supervisor.

- Thomas

jburk

  • Administrator
  • *****
  • Posts: 479
Re: supervisor_flags remove_logs
« Reply #1 on: May 03, 2010, 06:14:54 PM »
hmm... what version of Qube is running on your supervisor?  It may no longer be the default behavior.

Can you run 'qbadmin s --config', and post the line that contains "supervisor_flags"?  It will be set to an integer, but I can look it up against the bitmask and will tell you what it translates to.

Achilles

  • Sr. Member
  • ****
  • Posts: 25
Re: supervisor_flags remove_logs
« Reply #2 on: May 03, 2010, 07:11:19 PM »
hmm.. after strange things happend i moved the supervisor from Mac OS X10.5 to a Centos 5.4 host.

On OS X in /var/log/supelog no supervisor messages except "permission granted to user [xyz]" were logged. then after a while the supervisor would not distribute the work anymore to all the nodes. On the pipelinefx homepage i'm reading that 10.5 should be supported - so I'm a bit puzzled. What version of qube is needed for OS X 10.5 - the 10.3 or the 10.6 package?

After moving the supervisor to CentOS 5.4 i can see the following startup messaegs (and the other loglines too):

Code: [Select]
supervisor_flags: host_recontact,heartbeat_monitor,running_monitor,stub_optimize,retry_busy (7174)

so, it seems the docs should be updated to reflect the new defaults.

- Thomas