RE: configuration patch - final version

[Jan-Henrik Haukeland]

Well this was cool, Martin! I was thinking about doing it first this weekend,
but since you have already fixed it I can just sit back and relax, thanks :-)

So all our remaining tasks are done according to the monit 3.2 release plan,
thanks to Martin and Christian's hacker "tour de force". I propose that we test
the system a few days and release it this Friday, if everything is okay. Agree?

Ps.

I was thinking about Olivier's proposal about writing the state to a
"state-file". I suggest that we simply write the processlist to a binary file
(on reload and for each poll cycle) and upon reload or restart (in case of an
error) we re-read this file into a temporary processlist data-structure -
traverse the old list and for all processes that match the new list use the old
values especially reuse the do_validate flag per process. It's a very simple
but effective strategy and will solve the state problem. What do you think?

Jan-Henrik



> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden Behalf Of
> Martin Pala
> Sent: 11. februar 2003 18:19
> To: address@hidden
> Subject: configuration patch - final version
>
>
> There's final version of the patch, it implements:
>
> 1.) let monit reload only on SIGHUP
> 2.) new monit command 'reload' added - it will cause monit's
> reinitialization (send SIGHUP to running daemon)
> 3.) new monit option '-t' added - instructs monit to run in testing mode
> (just parse configuration and exit). Any error is reported by parser,
> otherwise success is signalized on terminal.
>
> Martin
>
> P.S. sorry for loading the list by so many patches - i remembered that
> we decided to add 'reload' command and '-t' option too after i send
> first patch
>