You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Chuck Murcko <ch...@n2k.com> on 1997/01/09 00:05:13 UTC
n + 1 MaxClients?
Anyone else seen MaxClients + 1 servers indicated in their status output?
I had 256 configured for awhile today, and was seeing things top out at
257 children. Now I'm configured for HARD_SERVER_LIMIT - 1, to be safe. 8^)
chuck
Chuck Murcko N2K Inc. Wayne PA chuck@telebase.com
And now, on a lighter note:
"You'll never be the man your mother was!"
Re: n + 1 MaxClients?
Posted by James H Cloos Jr <cl...@jhcloos.com>.
Chuck Murcko <ch...@n2k.com> writes:
> Anyone else seen MaxClients + 1 servers indicated in their status output?
I'd forgotten about this. We used to get this in 1.1.1, but we've not
been red-lined since I cut over to the 1.2bs.
-JimC
--
James H. Cloos, Jr. <URL:http://www.jhcloos.com/~cloos/>
cloos@jhcloos.com Work: cloos@io.com
LPF,Usenix,SAGE,ISOC,ACLU FB C2 B3 7A 50 44 AE 2D C7 4C 1E C1 EF 7A FB 30
Re: n + 1 MaxClients?
Posted by Ed Korthof <ed...@organic.com>.
On Jan 8, 3:53pm, Ed Korthof wrote:
> Subject: Re: n + 1 MaxClients?
> Looking at the code, I don't think that you actually had more than
> HARD_SERVER_LIMIT -- probably the a child switched from idle to busy status
> while the functions counted the children were running.
I should ammend that -- this holds only if you used the /status handler to get
the information, as the description implied. If 'ps -ef | grep httpd | grep -v
grep | wc -l' produces a number one larger than HARD_SERVER_LIMIT, that is
correct (since the parent server isn't counted amoung the children). If it
produces a number two larger, then there is a more serious problem and there
really are more than HARD_SERVER_LIMIT children running (unless there's
recently been a graceful restart, and some of the children are hanging around).
That would be very bad. I attempted to reproduce it, but was unable; if you
can produce that error, I'd like to know.
--
-- Ed Korthof | Web Server Engineer --
-- ed@organic.com | Organic Online, Inc --
-- (415) 278-5676 | Fax: (415) 284-6891 --
Re: n + 1 MaxClients?
Posted by Brian Behlendorf <br...@organic.com>.
Ah, cool, you addressed those concerns. +1 on these bugfix patches.
Brian
> However, there is one incorrect comparison (which still doesn't allow for more
> children than HARD_SERVER_LIMIT, though it would allow one more child than the
> maximum specified, if that was less than HARD_SERVER_LIMIT). There is also
> the possibility of messing up your server if you specify StartServers >
> HARD_SERVER_LIMIT (the latter may cause fatal memory corruption -- it did in
> the one test I've run). The attached patch fixes both of those.
>
> Ed
>
> --
> -- Ed Korthof | Web Server Engineer --
> -- ed@organic.com | Organic Online, Inc --
> -- (415) 278-5676 | Fax: (415) 284-6891 --
>
--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com www.apache.org hyperreal.com http://www.organic.com/JOBS
Re: n + 1 MaxClients?
Posted by Ed Korthof <ed...@organic.com>.
On Jan 8, 6:05pm, Chuck Murcko wrote:
> Subject: n + 1 MaxClients?
> Anyone else seen MaxClients + 1 servers indicated in their status output?
>
> I had 256 configured for awhile today, and was seeing things top out at
> 257 children. Now I'm configured for HARD_SERVER_LIMIT - 1, to be safe. 8^)
>
> chuck
> Chuck Murcko N2K Inc. Wayne PA chuck@telebase.com
> And now, on a lighter note:
> "You'll never be the man your mother was!"
>-- End of excerpt from Chuck Murcko
Looking at the code, I don't think that you actually had more than
HARD_SERVER_LIMIT -- probably the a child switched from idle to busy status
while the functions counted the children were running.
However, there is one incorrect comparison (which still doesn't allow for more
children than HARD_SERVER_LIMIT, though it would allow one more child than the
maximum specified, if that was less than HARD_SERVER_LIMIT). There is also
the possibility of messing up your server if you specify StartServers >
HARD_SERVER_LIMIT (the latter may cause fatal memory corruption -- it did in
the one test I've run). The attached patch fixes both of those.
Ed
--
-- Ed Korthof | Web Server Engineer --
-- ed@organic.com | Organic Online, Inc --
-- (415) 278-5676 | Fax: (415) 284-6891 --
Re: n + 1 MaxClients?
Posted by Rob Hartill <ro...@imdb.com>.
On Wed, 8 Jan 1997, Chuck Murcko wrote:
> Anyone else seen MaxClients + 1 servers indicated in their status output?
I reported that a long time ago. Back in the days before the scoreboard
monitor was written and I had a perl version.
I thought I'd fixed it, or maybe nobody cared at the time, or maybe it's
a copycat bug.