|
From: | Martin Pala |
Subject: | Re: device stuff - cvs |
Date: | Mon, 09 Jun 2003 23:36:22 +0200 |
User-agent: | Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3.1) Gecko/20030527 Debian/1.3.1-2 |
Jan-Henrik Haukeland wrote:
I'm going to holiday on monday for 14 days, so i hope everything will be OK and new monit release will be ready :)We'll try to manage and release something before you get back, but wewill miss you!I look forward to check out the device stuff, (I haven't got around to it yet, but I'll check out your patch later this evening). I agree with your previous mail where you state, INCOMPATIBLE CHANGES IN COMPARISION WITH MONIT <= 3.x:. and changing the CHECK-statement to CHECK {DEVICE|DIRECTORY|FILE|PROCESS} service_name PATH /depends/on/context this is fine. But I hope that we can avoid breaking backward compability and still use CHECK service_name PATH /depends/on/context for ordinary processes. I'll try to check out if it's possible to do some grammar stunts in the parser when I look at the patch.
Its done now (100% backward compatibility kept).I'm going holiday in few hours - i'm sorry i didn't managed to finish all stuff i wanted to do. Remaining items:
- *BSD + HPUX + AIX + MACOSX port. Maybe it will work on some of these architectures nut it is tested just on Solaris and Linux. Filesystem stuff will be probably non functional on *BSD which has different interface. This should be done before we release 4.0.
- size test for files and directories - it is not critical, we can add it after 4.0
- owner (uid) and group (gid) test for files and directories - it is not critical, we can add it after 4.0
I though yet about one topic - it is possible to consolidate syntax of resource checks with syntax of device style, for example:
"if cpuusage is greater than 60.0 for 2 cycles then alert" -> "if cpu usage > 60 % for 2 cycles then alert"
"if memkbyte > 100000.0 for 5 cycles then stop" -> "if memory usage > 100 MB for 5 cycles then stop"
"if memusage > 10.0 for 5 cycles then stop" -> "if memory usage > 10 % for 5 cycles then stop"
It is probably more simple i think (it is sufficient to have just 'cpu' and 'mem(ory)' statements instead of present 'cpuusage', 'memkbyte' and 'memusage'). What about it?
I'll be back on 20.6. - i'm not sure wheter i'll have internet connection (i plan to do some trips and let the work sleep :)
Cheers, Martin
[Prev in Thread] | Current Thread | [Next in Thread] |