Uploaded image for project: 'ZABBIX BUGS AND ISSUES'
  1. ZABBIX BUGS AND ISSUES
  2. ZBX-13062

Under Windows, Zabbix agent will exit, if timing out at the exact right moment.

    XMLWordPrintable

Details

    • Team C
    • Sprint 22, Sprint 23, Sprint 26, Sprint 27, Sprint 28, Sprint 29
    • 1

    Description

      Testing item vfs.dir.count[] under Windows, it was possible to cause Zabbix agent to hang or exit. The last lines in log are:

      2236:20171117:153542.878 vfs_dir_count() pro=2 count=30
      2236:20171117:153542.893 vfs_dir_count() 10=c:\/Z/zabbix-4.0.0alpha1=1 tmA/C/M=1510312786/1510312715/1510312712=? size=0=OK
      2236:20171117:153542.910 vfs_dir_count() pro=2 count=31
      2276:20171117:153542.925 WIN zbx_execute_threaded_metric() TIMEOUTED: WaitForSingleObject()
      zabbix_agentd.exe [2276]: ERROR file:..\..\..\src\libs\zbxsys\mutexs.c,line:135 Something impossible has just happened.

      To reproduce, one of the threaded metrics (vfs.dir.size, vfs.dir.count, vfs.fs.size) must be writing something into zabbix_agentd.log at the moment they time out. That is a very slim possibility, although the first two items are naturally long-running.

      To reproduce artificially, void __zbx_zabbix_log(int level, const char *fmt, ...) can be slowed down with e.g. Sleep(1).

      As per https://msdn.microsoft.com/en-us/library/windows/desktop/ms687032(v=vs.85).aspx , WAIT_ABANDONED is returned when child process failed to release mutex before termination. Zabbix agent chooses to terminate in response.

      Expected:
      Zabbix agent should not terminate under unusual conditions that do not themselves prevent its further function.

      Attachments

        Issue Links

          Activity

            People

              MVekslers Michael Veksler
              valdis Valdis Kauķis (Inactive)
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: