You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Rodent of Unusual Size <co...@decus.org> on 1997/07/26 05:03:49 UTC
Something's rotten in the state of 1.3a2-dev..?
I sent this earlier to-day, but it doesn't seem to have made it to
the list - hence my last query about whether the list was working..
#ken :-/}
END
{just in case the message got caught by administrivia last time}
------------------------------------------------------------------------
Date: Fri, 25 Jul 1997 16:25:45 -0400
From: coar@decus.org (Rodent of Unusual Size)
Subject: Something's rotten in the state of 1.3a2-dev
Bad news, confreres.. I've just rebuilt from the latest stuff in
HEAD, and two new problems seem to have been introduced:
1. A SIGHUP is now giving me
bind: Address already in use
httpd: could not bind to port 9000
I have NO Listen directives, and only one IP address for
the host in question. All vhosts are name-based. This is OSF/1
V3.2 on an Alpha, *not* Solaris.
2. The server never responds. A simple HEAD request nets me
nothing; no "HTTP/1.1 200 OK", nothing. It doesn't even respond
if I give it bogus HTTP. The server's gone off into la-la land,
and I lack Dean's winning ways with performance tools to tell me
what it's doing. It's not mod_rewrite because I've taken that
out; it's not mod_mime_magic because I never put it in. I doubt
it's the 64-bit stuff even though this is an Alpha. It's not
the mod_include changes, either, 'cuz that's commented out now,
too. I strongly suspect the CGI-unbuffering stuff, since that
touched an awful lot of Deep Magic modules..
Is the latest stuff in HEAD working for anyone else? Dean, do you
have any OSF tool invocations I can try?
#ken :-(}
Re: Something's rotten in the state of 1.3a2-dev..?
Posted by Dean Gaudet <dg...@arctic.org>.
www.arctic.org has been running HEAD as of all those changes I committed
including the cgi stuff and yadda yadda on wednesday.
I'll go see if I can reproduce it on my boxes. If not I'll try under
OSF/1 sometime this weekend.
Dean
On Fri, 25 Jul 1997, Rodent of Unusual Size wrote:
> I sent this earlier to-day, but it doesn't seem to have made it to
> the list - hence my last query about whether the list was working..
>
> #ken :-/}
> END
> {just in case the message got caught by administrivia last time}
> ------------------------------------------------------------------------
> Date: Fri, 25 Jul 1997 16:25:45 -0400
> From: coar@decus.org (Rodent of Unusual Size)
> Subject: Something's rotten in the state of 1.3a2-dev
>
> Bad news, confreres.. I've just rebuilt from the latest stuff in
> HEAD, and two new problems seem to have been introduced:
>
> 1. A SIGHUP is now giving me
>
> bind: Address already in use
> httpd: could not bind to port 9000
>
> I have NO Listen directives, and only one IP address for
> the host in question. All vhosts are name-based. This is OSF/1
> V3.2 on an Alpha, *not* Solaris.
>
> 2. The server never responds. A simple HEAD request nets me
> nothing; no "HTTP/1.1 200 OK", nothing. It doesn't even respond
> if I give it bogus HTTP. The server's gone off into la-la land,
> and I lack Dean's winning ways with performance tools to tell me
> what it's doing. It's not mod_rewrite because I've taken that
> out; it's not mod_mime_magic because I never put it in. I doubt
> it's the 64-bit stuff even though this is an Alpha. It's not
> the mod_include changes, either, 'cuz that's commented out now,
> too. I strongly suspect the CGI-unbuffering stuff, since that
> touched an awful lot of Deep Magic modules..
>
> Is the latest stuff in HEAD working for anyone else? Dean, do you
> have any OSF tool invocations I can try?
>
> #ken :-(}
>
Re: Something's rotten in the state of 1.3a2-dev..?
Posted by Dean Gaudet <dg...@arctic.org>.
Neither of these are happening for me under linux/x86. Try a complete
rebuild, just in case you've got some cruft hanging around. If still no
go I'll try OSF/1 tomorrow.
I ran the 64-bit patch on a linux/alpha box, but maybe I still messed it
up.
Dean
On Fri, 25 Jul 1997, Rodent of Unusual Size wrote:
> I sent this earlier to-day, but it doesn't seem to have made it to
> the list - hence my last query about whether the list was working..
>
> #ken :-/}
> END
> {just in case the message got caught by administrivia last time}
> ------------------------------------------------------------------------
> Date: Fri, 25 Jul 1997 16:25:45 -0400
> From: coar@decus.org (Rodent of Unusual Size)
> Subject: Something's rotten in the state of 1.3a2-dev
>
> Bad news, confreres.. I've just rebuilt from the latest stuff in
> HEAD, and two new problems seem to have been introduced:
>
> 1. A SIGHUP is now giving me
>
> bind: Address already in use
> httpd: could not bind to port 9000
>
> I have NO Listen directives, and only one IP address for
> the host in question. All vhosts are name-based. This is OSF/1
> V3.2 on an Alpha, *not* Solaris.
>
> 2. The server never responds. A simple HEAD request nets me
> nothing; no "HTTP/1.1 200 OK", nothing. It doesn't even respond
> if I give it bogus HTTP. The server's gone off into la-la land,
> and I lack Dean's winning ways with performance tools to tell me
> what it's doing. It's not mod_rewrite because I've taken that
> out; it's not mod_mime_magic because I never put it in. I doubt
> it's the 64-bit stuff even though this is an Alpha. It's not
> the mod_include changes, either, 'cuz that's commented out now,
> too. I strongly suspect the CGI-unbuffering stuff, since that
> touched an awful lot of Deep Magic modules..
>
> Is the latest stuff in HEAD working for anyone else? Dean, do you
> have any OSF tool invocations I can try?
>
> #ken :-(}
>
Re: Something's rotten in the state of 1.3a2-dev..?
Posted by Marc Slemko <ma...@worldgate.com>.
On Fri, 25 Jul 1997, Rodent of Unusual Size wrote:
> I sent this earlier to-day, but it doesn't seem to have made it to
> the list - hence my last query about whether the list was working..
>
> #ken :-/}
> END
> {just in case the message got caught by administrivia last time}
> ------------------------------------------------------------------------
> Date: Fri, 25 Jul 1997 16:25:45 -0400
> From: coar@decus.org (Rodent of Unusual Size)
> Subject: Something's rotten in the state of 1.3a2-dev
>
> Bad news, confreres.. I've just rebuilt from the latest stuff in
> HEAD, and two new problems seem to have been introduced:
>
> 1. A SIGHUP is now giving me
>
> bind: Address already in use
> httpd: could not bind to port 9000
>
> I have NO Listen directives, and only one IP address for
> the host in question. All vhosts are name-based. This is OSF/1
> V3.2 on an Alpha, *not* Solaris.
What does NO_SLACK do?
> 2. The server never responds. A simple HEAD request nets me
> nothing; no "HTTP/1.1 200 OK", nothing. It doesn't even respond
> if I give it bogus HTTP. The server's gone off into la-la land,
> and I lack Dean's winning ways with performance tools to tell me
> what it's doing. It's not mod_rewrite because I've taken that
> out; it's not mod_mime_magic because I never put it in. I doubt
> it's the 64-bit stuff even though this is an Alpha. It's not
> the mod_include changes, either, 'cuz that's commented out now,
> too. I strongly suspect the CGI-unbuffering stuff, since that
> touched an awful lot of Deep Magic modules..
Are the child processes running? Do a gdb on a child, or if you have
anything ktrace or truss-like use it.
Try checking out a tree from a week ago, then narrowing down exactly when
the problem started.
Re: Something's rotten in the state of 1.3a2-dev..?
Posted by Paul Sutton <pa...@ukweb.com>.
On Fri, 25 Jul 1997, Rodent of Unusual Size wrote:
> 2. The server never responds. A simple HEAD request nets me
> nothing; no "HTTP/1.1 200 OK", nothing. It doesn't even respond
> if I give it bogus HTTP. The server's gone off into la-la land,
> and I lack Dean's winning ways with performance tools to tell me
> what it's doing. It's not mod_rewrite because I've taken that
> out; it's not mod_mime_magic because I never put it in. I doubt
> it's the 64-bit stuff even though this is an Alpha. It's not
> the mod_include changes, either, 'cuz that's commented out now,
> too. I strongly suspect the CGI-unbuffering stuff, since that
> touched an awful lot of Deep Magic modules..
If you compiled with -O, try compiling without. I had a problem on OSF 3.2
where a certifiably correct and exitable loop caused an infinite loop on
OSF when compiled with -O. It worked fine without -O or with -g. I.e. was
an optimiser bug.
//pcs