Re: SIGSEGV

[Jan-Henrik Haukeland]

Christian Hopp <address@hidden> writes:

> On Mon, 17 Feb 2003, Jan-Henrik Haukeland wrote:
> 
> >
> > Ahh, I got a SIGSEGV when testing monit now in association with monit
> > reload. I'll have to nail this down before any release.
> 
> Strange I dont have a problem... only if I bombard it with too many
> SIGHUPs at a time.  Then I had a SIGSEGV and I get some "monit: No daemon
> process found" from time to time.

That's the same stuff I experienced only it happened when I did *one*
reload (maybe you start monit from init, so it respawns before the
problem manifests itself from userland). Anyway the problem is fixed
(I hope) and described in the latest cvs check in. (On a side note, if
a monit daemon dies without any log messages it's almost certainly
because off a SIGSEGV or another crash signal)

> But everytime I do a "SIGHUP" or "monit reload" I get the following log
> message:
> 
> monit: Openssl engine error during update_ssl_cert_data():
>        error:00000000:lib(0):func(0):reason(0)
> monit: The client did not supply a required client certificate!
> http server: Cannot establish SSL connection
> 
> But who is connecting to the server during a SIGHUP????

The monit daemon :-) Becasue a signal handler is asynchronous. But
reload on sighup is now synchronous.

-- 
Jan-Henrik Haukeland