Re: check file timestamp bug?

[Jan-Henrik Haukeland]

On 21. sep. 2007, at 04.23, Alex Stewart wrote:

> I am having a problem with monit that occurs in both v4.8.1 and v4.9
> and the changelog (http://www.tildeslash.com/monit/CHANGES.txt)
> appears to specifically state that the bug was fixed in v4.8:
> ("Removed a feature introduced in 4.7 which tested that a check-file,
> check-directory or check-fifo actually refered to an existing object
> of that type.  Monit should not require these file objects to exist at
> startup.")

This refer to paths in a check entry, e.g. you can write

check file testMat_pidfile with path /var/run/testMat.pid

without monit complaining at startup if /var/run/testMat.pid does not  
exist. With regards to the timestamp test monit expect this file to  
exist so it can take an initial timestamp at startup and compare the  
file stamp in subsequent runs. However I can see your point when  
monitoring in non-active mode. I have changed the monit code in CVS  
to allow the "if changed timestamp" test to refer to a non-existing  
file so we are consistent. Note that this mean the initial timestamp  
will be 0 and as long as the file does not exist monit will issue an  
alert (in active mode). As soon as the file exist, the test "if  
changed timestamp" will be true and execute its action. The timestamp  
is saved and the next time the test run it will work as expected.

> Given the wording of the changelog for v4.8 - and what would seem to
> be logical behaviour (why should monit check for files that are mode
> manual until the group has been started?) this would appear to be a
> bug.

The change will be part of the next monit release or you may check it  
out from CVS[1] now if you want to test.

Jan-Henrik

[1] Use these instructions to check out from CVS http:// 
savannah.nongnu.org/cvs/?group=monit