Hi, I am facing the problem with monit and heartbeat and server crashes, so that monit starts after crash-reboot manual services when it looks up it's state file. In the archives it was fist mentioned by Marco Bisioli in September 2003. To resolve this issue, the use of a ramdisk or a cleanup-startscript was suggested.
I would be glad if monit could do this. Here is one (hopefully not too stupid) suggestion: As far as I understand the concept, when monit unexpectedly died and is restarted ( for example by init) then it should take a look at it's state table and control the services as it did before it died. But when the server crashes and monit and all the controlled services die, monit should not take a look at it's state table at system startup, because then it happens that manual controlled services which were previously started, are restarted, although this might not be wanted because heartbeat started them on the other node. So couldn't monit when it starts and detects it's old state file, compare the age of the file with the uptime of the server ? If the uptime is smaller than the age of the state file then monit has to assume that the server crashed and ignores the old state file. If the uptime is greater than the age of the state file then monit has to assume that is was normally restarted manually or by init after it's own crash and accepts the old state file like it does at the moment. Regards Tim Einmahl -- To unsubscribe: http://lists.nongnu.org/mailman/listinfo/monit-general
