1) this has been discussed already but I have this problem again with flapping monitors which trigger threshold several times.
Please implement new variable - current trigger condition, so based on tha variable you can change theshold of the trigger with hysteresis. (To make trigger ON condition treshold to be higher than trigger OFF condition, this will completely resolve trigger flap and page storm problem).
2) Please make zabbix-server to detect stale PID file (check if process listed in PID actually exists), and to not crash when database when down - it should probably hang there and wait for database to go up.
My zabbix server is down pretty often (didnt start after reboot or died when somebody restarted DB). I have to write some scripts to keep it up which is probably not a good way of doing so.
3) I have problem with the queue, i have 50 server daemons runnign but if i have 5 hosts down queue gets completely backed up. SOmething is wrong there in queue algorythm, it probably shouldnt use more than 5 daemons to sit there and waiting for timeout to query those 5 servers that are down (the rest 45 daemons should do the job processing the resto of the queue) but somehow it is enough to make complete queue backup.
Please implement new variable - current trigger condition, so based on tha variable you can change theshold of the trigger with hysteresis. (To make trigger ON condition treshold to be higher than trigger OFF condition, this will completely resolve trigger flap and page storm problem).
2) Please make zabbix-server to detect stale PID file (check if process listed in PID actually exists), and to not crash when database when down - it should probably hang there and wait for database to go up.
My zabbix server is down pretty often (didnt start after reboot or died when somebody restarted DB). I have to write some scripts to keep it up which is probably not a good way of doing so.
3) I have problem with the queue, i have 50 server daemons runnign but if i have 5 hosts down queue gets completely backed up. SOmething is wrong there in queue algorythm, it probably shouldnt use more than 5 daemons to sit there and waiting for timeout to query those 5 servers that are down (the rest 45 daemons should do the job processing the resto of the queue) but somehow it is enough to make complete queue backup.
Comment