On 2.0 again.

[rb...@covalent.net]

We are back up on 2.0.  I think we were down for about thirty seconds
while I made the move over.  Brian, is there a chance you could look into
why we moved back to 1.3 at 00:00:01 last night?  I am available on my
cell phone if you have any questions about what I did to turn it on/off.

Ryan


_______________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------

Re: On 2.0 again.

[Greg Ames <gr...@raleigh.ibm.com>]

Brian Behlendorf wrote:
> 
> On Fri, 16 Feb 2001 rbb@covalent.net wrote:
> > We are back up on 2.0.  
> 
> Might as well leave it running!  
> 
> Would be nice to find a good summary of the threading issues we could
> forward to the freebsd-stable list, unless it's just "general
> instability".
> 
The biggest problem I know about is the loop when we just enable
threads.  I can't tell you exactly what Apache is doing to trigger this,
but it's easy to recreate with log replay.    

Here's what Jeff found by attaching gdb to one of the 3 looping
processes.  I did post this before but maybe you didn't notice it.

[...many "Reading symbols from blah" deleted]

Reading symbols from /usr/libexec/ld-elf.so.1...done.
0x2813f1e2 in _thread_sigframe_restore () from /usr/lib/libc_r.so.4
(gdb)

Greg

Re: On 2.0 again.

[rb...@covalent.net]

> > We are back up on 2.0.  I think we were down for about thirty seconds
> > while I made the move over.  Brian, is there a chance you could look into
> > why we moved back to 1.3 at 00:00:01 last night?  I am available on my
> > cell phone if you have any questions about what I did to turn it on/off.
> 
> I edited /usr/local/adm/logrotate to call /usr/local/apache2/bin/apachectl
> graceful instead of /usr/local/apache/bin/apachectl, and at midnight it
> rotated the log as usual.  Cool!
> 
> I will be online tomorrow until the evening, but offline Sunday.  Might as
> well leave it running!  Sorry I killed the wrong one last night, I
> presumed you had taken it down again intentionally.

I'm on "vacation" through Tuesday, but I'll be on e-mail once in a while
this weekend.  If there are any REALLY important questions, I'll reply.

> Would be nice to find a good summary of the threading issues we could
> forward to the freebsd-stable list, unless it's just "general
> instability".

It's general instability.  The current list looks something like:

Sendfile just doesn't work at all
threads go off and loop forever, and we don't know where.

Ryan

_______________________________________________________________________________
Ryan Bloom                        	rbb@apache.org
406 29th St.
San Francisco, CA 94131
-------------------------------------------------------------------------------

Re: On 2.0 again.

[Brian Behlendorf <br...@collab.net>]

On Fri, 16 Feb 2001 rbb@covalent.net wrote:
> We are back up on 2.0.  I think we were down for about thirty seconds
> while I made the move over.  Brian, is there a chance you could look into
> why we moved back to 1.3 at 00:00:01 last night?  I am available on my
> cell phone if you have any questions about what I did to turn it on/off.

I edited /usr/local/adm/logrotate to call /usr/local/apache2/bin/apachectl
graceful instead of /usr/local/apache/bin/apachectl, and at midnight it
rotated the log as usual.  Cool!

I will be online tomorrow until the evening, but offline Sunday.  Might as
well leave it running!  Sorry I killed the wrong one last night, I
presumed you had taken it down again intentionally.

Would be nice to find a good summary of the threading issues we could
forward to the freebsd-stable list, unless it's just "general
instability".

	Brian