Re: start and stop process by monit

[Kenneth Yip]

I believe it is fine to have it as an option so that user can choose whether to
send SIGKILL to
a process. Of course, if you use monit to monitor a database, you are not going
to
send a SIGKILL to it, however, for web servers which sometimes hung up, it is
good
to have SIGKILL.

On the other hand, refining the restart process seems to be even better. In the
current version,
if the stop process timeout, the monitoring of the process failed. It would be
great if  it is possible to find tune this, such as giving more tolerence to
monit  like
    stop_timeout in 3 cycles
    send SIGKILL if stop_timeout or stop again if stop_timeout
Especially on busy server, the stopping of a process may take much longer time
than expected. Rather than leaving the process unmonitored, it is better to have
monit
continues the monitoring and stop/restart the process in question.

Also, for tcp services, is it possible to add some tolerence to it? For example,
HTTP
may failed due to server busy for 1 cycle and may resume normal on the next poll
cycle without intervention. Adding option such as 'if 3 failed http in 3 cycle
than restart' may avoid
restart that is not necessary to recover a service.

Just my 2 cents.

Regards,
Kenneth Yip

----- Original Message ----- 
From: "Martin" <address@hidden>
To: "This is the general mailing list for monit" <address@hidden>
Sent: Tuesday, October 07, 2003 2:55 PM
Subject: Re: start and stop process by monit


> It is very dangerous to kill process which is not stopping within constant
timeframe - in the case of database which is for example handling logs you can
cause lot of problems (at minimum as soon as you will attempt to start it in the
future, you will wait for recovery to finish, which can take very long) => i
think it should be optional and that it is sufficient to kill the process in
stop method if the user wish it.
>
> -1 for adding killing the process from monit
>
> Martin
>
>
>
> ----- Původní zpráva -----
> Od: Jan-Henrik Haukeland <address@hidden>
> Datum: Úterý, 7.října 2003 - 0:13 dop.
> Předmět: Re: start and stop process by monit
>
> > Jan-Henrik Haukeland <address@hidden> writes:
> >
> > > If you really want to make sure that a process was stopped, you
> > can in
> > > the stop program or in the script monit calls to stop the
> > process, add
> > > code that kills the process with SIGKILL if graceful termination did
> > > not succeed. (Of course, this could instead be implemented in monit,
> > > i.e. if monit failed to stop a process it sends SIGKILL to the
> > process> as it's final action, but I think this decision should be
> > done in the
> > > stop program monit calls and not by monit).
> >
> > On second thought, maybe monit *should* send a process the SIGKILL
> > signal as its last action if the process wont stop after monit called
> > the stop program for the process. What do others think?
> >
> > -- 
> > Jan-Henrik Haukeland
> >
> >
> > --
> > To unsubscribe:
> > http://mail.nongnu.org/mailman/listinfo/monit-general
> >
>
>
>
>
>
> --
> To unsubscribe:
> http://mail.nongnu.org/mailman/listinfo/monit-general
>