Opened 3 years ago

Closed 17 months ago

#1131 closed defect (fixed)

munin-cgi-graph can "hang" due to a previous crash

Reported by: purewhite Owned by: nobody
Priority: normal Milestone: Munin 1.4.7
Component: web-interface Version: 1.4.5
Severity: normal Keywords: munin-cgi-graph, semaphore
Cc:

Description

Due to semaphores, the perl cgi can sit waiting forever if a previous crash doesn't clear the semaphores.
A server (apache) crash meant the semaphores didn't get cleared, even though the processes had ended. I don't understand semaphores enough to know why they didn't get cleared, as my understanding is that they should get cleared if the process dies.
Clearing the semaphore manually with ipcs and ipcrm allowed munin-cgi-graph to work again. Previously it would "timeout".

#814 seems to be on the right track with SEM_UNDO. Not sure of other methods to make this work.

Change History (1)

comment:1 Changed 17 months ago by snide

  • Resolution set to fixed
  • Status changed from new to closed

The 2.0.x line doesnt use any semaphores, but relies on a fastcgi container.
This should fix the bug.

Note: See TracTickets for help on using tickets.