for those who would like to try the latest pre-release of mon 1.0.0
(tagged "mon-1-0-0pre3" in CVS), you may obtain it here:
ftp://ftp.kernel.org/pub/software/admin/mon/devel/
the most important update is the consolidation of the monitor exit
processing and the trap/traptimeout handling code which has resulted
in more sensible code and a bunch of bug fixes with respect to trap and
traptimeout handling.
this should be paired with the mon-client-1.0.0pre2 perl module (there
have been no updates to that since 1.0.0pre2).
Changes between mon-1.0.0pre1 and mon-1.0.0pre3
Mon Jul 12 09:12:29 EDT 2004
---
-changed README to refer to the new, more sensible name for the perl module
client, which is mon-client
-applied eric's updates to INSTALL and added a mention of monshow and mon.cgi as
the web interfaces
-added eric's rpm spec file (i removed the patches because they are no longer
needed)
-added lmb's syslog.monitor (a nifty hack)
-added 'alertevery strict' code and docs, updated the README and INSTALL to
mention CVS, updated CREDITS
-incorporated mon.cgi 1.52
-minor addition to alert behavior explanation in mon.8
-in dialin.monitor.wrap.c, return the exit status of execv (if it fails, that is)
-fixed path to perl in file_change.monitor and smtp3.monitor
-added some rcs tags to identify the file versions
-handle_trap_timeout now calls process_event, and it works fine with
alert/upalert/alertevery/etc. as shown by my testing
-received traps now reset the trap timeout counter, and fixed some
other stuff wrt trap timeouts
-added sub process_event and made proc_cleanup and handle_trap use it
so that the alert mgmt code is shared rather than in two places. i tested
as much of it as i could and all seems to work well now, especially
upalert, alertafter, alertevery with traps.
-added per-service "_monitor_duration" variable which records how many
seconds the previous monitor took to execute. this is available via
"list opstatus". if no monitor has executed yet then the value is -1.
-added per-service "_monitor_running" variable whose value is 0 or 1
depending on whether the monitor is currently running for that service.
-removed gunk from handle_trap regarding the various TRAP_COLDSTART, etc.
processing, since most of it was a bad idea anyway, or at least as far as
i could tell. traps and their exit values are now processed exactly as
monitors are, which simplifies things greatly and adds to more intuitive
functionality. this means the "spc" value in a trap is now ignored.
-fixed some args processing in call_alert
-fixed a bug which would prevent alerts or upalerts
from being sent when call alerts is passed the "output"
argument whose value is undef
-remove usage of parse_line in trap processing
(backported from mon 1.1 code)
-make esc_str escape spaces in order to be compatible with monperl-1-0-0pre1
-added list of all possible client commands to moncmd
-added --community to set the snmp community in reboot.monitor
-patch to traceroute.monitor from meekj
added StateDir, TracerouteOptions, StopAt config options
some bugfixes to config file parsing
reap children to avoid defunct processes
added timeout alarm
-up_rtt.monitor
added -r to log individual rtts, better error reporting for tcp and udp check
___
mon mailing list
[EMAIL PROTECTED]
http://linux.kernel.org/mailman/listinfo/mon