You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Cliff Woolley <cl...@yahoo.com> on 2001/07/10 05:28:06 UTC
Re: Fixes to 2.0.20
On Mon, 9 Jul 2001 rbb@covalent.net wrote:
> I have re-rolled, and I will move the new 2.0.20 into
> http://dev.apache.org/dist in a few minutes.
I'm still seeing some very slightly odd things with graceful restart on
the threaded MPM, but overall it seems to work fairly well and passes the
(non-tortuous) tests I threw at it.
+1 for beta
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA
Re: Fixes to 2.0.20
Posted by Cliff Woolley <cl...@yahoo.com>.
On Tue, 10 Jul 2001, Greg Ames wrote:
> > Then do a graceful restart.
> > You can see the server shifting itself around as the new processes come up
> > and the old ones go down... but for some reason three workers with
> > generations 3, 4, and 5 respectively get left behind, one per "line" in
> > the server-status output (whatever that means),
>
> Once upon a time, all the threads for a given process were displayed on a
> single line if possible. ISTR the leftmost position on the line was unused,
> possibly to reserve space for process status.
On one try, the three 'G's (one per line) were in the middle of the line.
On the next two tries, they were all in the leftmost position of their
lines. <shrug>
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA
Re: Fixes to 2.0.20
Posted by Greg Ames <gr...@remulak.net>.
On Tuesday 10 July 2001 00:59, Cliff Woolley wrote:
> Then do a graceful restart.
> You can see the server shifting itself around as the new processes come up
> and the old ones go down... but for some reason three workers with
> generations 3, 4, and 5 respectively get left behind, one per "line" in
> the server-status output (whatever that means),
Once upon a time, all the threads for a given process were displayed on a
single line if possible. ISTR the leftmost position on the line was unused,
possibly to reserve space for process status.
Greg
Re: Fixes to 2.0.20
Posted by Cliff Woolley <cl...@yahoo.com>.
On 10 Jul 2001, Jeff Trawick wrote:
> graceful can hang with prefork and threaded can segfault on idle
> server maintenance, both new problems since the previous beta AFAIK
>
> -1 for beta :(
Crap. :-( Yeah, that's a problem. Maybe we can fix these problems and
get a 2.0.21 rolled by mid-week.
Don't know if this information helps any, but here's what I was seeing on
graceful restart with threaded. Bring up 2.0.20 with the threaded MPM
with the default MPM parameters. Pull up a /server-status browser window
and set it to refresh every few seconds. Then do a graceful restart.
You can see the server shifting itself around as the new processes come up
and the old ones go down... but for some reason three workers with
generations 3, 4, and 5 respectively get left behind, one per "line" in
the server-status output (whatever that means), even though the actual
server's generation is now only 1. How they got to 3, 4, and 5, I don't
know. What's more, they don't go away even if you then hit the server
with 1000 requests using ab. A non-graceful restart took them out. I had
some difficulty at that point getting apachectl stop to work, but
eventually the server did go down without me having to do anything TOO
drastic.
I'll try to provide more details when I get a chance.
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA
Re: Fixes to 2.0.20
Posted by Jeff Trawick <tr...@attglobal.net>.
Cliff Woolley <cl...@yahoo.com> writes:
> On Mon, 9 Jul 2001 rbb@covalent.net wrote:
>
> > I have re-rolled, and I will move the new 2.0.20 into
> > http://dev.apache.org/dist in a few minutes.
>
> I'm still seeing some very slightly odd things with graceful restart on
> the threaded MPM, but overall it seems to work fairly well and passes the
> (non-tortuous) tests I threw at it.
>
> +1 for beta
graceful can hang with prefork and threaded can segfault on idle
server maintenance, both new problems since the previous beta AFAIK
-1 for beta :(
--
Jeff Trawick | trawick@attglobal.net | PGP public key at web site:
http://www.geocities.com/SiliconValley/Park/9289/
Born in Roswell... married an alien...