Finally: the crash goes away with this patch, which was the only thing I
could think of trying without digging further into the code to find out why
s->inf->cs_sum was null at that point (and why this problem affects me and
not anyone else):
--- validate.c.orig 2008-03-25 16:17:44.000000000 +0000
+++ validate.c 2008-03-25 16:18:03.000000000 +0000
@@ -985,11 +985,11 @@
/* if we are testing for changes only, the value is variable */
if(cs->test_changes) {
- Event_post(s, EVENT_CHANGED, STATE_FAILED, cs->action,
- "checksum was changed for %s", s->path);
/* reset expected value for next cycle */
FREE(cs->hash);
cs->hash= xstrdup(s->inf->cs_sum);
+ Event_post(s, EVENT_CHANGED, STATE_FAILED, cs->action,
+ "checksum was changed for %s", s->path);
} else {
/* we are testing constant value for failed or passed state */
Event_post(s, EVENT_CHECKSUM, STATE_FAILED, cs->action,
Unfortunately this isn't a solution, as the file monitoring status changes
to "not monitored" after one change to the file, and any further changes to
the file do not cause the daemon to be restarted. This of course could just
be user error, and maybe there's a correct way to do what I'm trying to do.
Regards,
Brian.
_______________________________________________
monit-dev mailing list
address@hidden
http://lists.nongnu.org/mailman/listinfo/monit-dev