You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Jim Jagielski <ji...@jaguNET.com> on 1997/12/12 15:05:34 UTC
child_timeouts and set_callback_and_alarm
This is mostly to Dean... I usually don't run with the status module
anymore but figured I'd give it a try. Thing is, with 1.3b3, the SS
value is way wrong... I _think_ it has something to do with the
way set_callback_and_alarm() is coded since timeout_len isn't
being adjusted (since child_timeouts is true)... Thus, _rtime isn't
being adjusted as time goes by :/
--
====================================================================
Jim Jagielski | jaguNET Access Services
jim@jaguNET.com | http://www.jaguNET.com/
"Look at me! I'm wearing a cardboard belt!"
Re: child_timeouts and set_callback_and_alarm
Posted by Dean Gaudet <dg...@arctic.org>.
Ok the solution is going to be to change a bunch of the #ifndef
OPTIMIZE_TIMEOUTS to also be compiled when OPTIMIZE_TIMEOUTS but also test
child_timeouts. Grr.
Dean
On Fri, 12 Dec 1997, Jim Jagielski wrote:
> This is mostly to Dean... I usually don't run with the status module
> anymore but figured I'd give it a try. Thing is, with 1.3b3, the SS
> value is way wrong... I _think_ it has something to do with the
> way set_callback_and_alarm() is coded since timeout_len isn't
> being adjusted (since child_timeouts is true)... Thus, _rtime isn't
> being adjusted as time goes by :/
> --
> ====================================================================
> Jim Jagielski | jaguNET Access Services
> jim@jaguNET.com | http://www.jaguNET.com/
> "Look at me! I'm wearing a cardboard belt!"
>
Re: child_timeouts and set_callback_and_alarm
Posted by Dean Gaudet <dg...@arctic.org>.
Yeah there is a race -- if the child can wake up, serve a request, then go
back to sleep before the parent does its child maintenance then last_rtime
won't be updated. But it has to keep doing this consistently... maybe I
just don't see it as much on my servers.
A solution is to add back the STATUS time field I removed in favour of
rtime, and set it in the child using r->request_time or something like
that to avoid the extra call to time().
Dean
On Fri, 12 Dec 1997, Jim Jagielski wrote:
> This is mostly to Dean... I usually don't run with the status module
> anymore but figured I'd give it a try. Thing is, with 1.3b3, the SS
> value is way wrong... I _think_ it has something to do with the
> way set_callback_and_alarm() is coded since timeout_len isn't
> being adjusted (since child_timeouts is true)... Thus, _rtime isn't
> being adjusted as time goes by :/
> --
> ====================================================================
> Jim Jagielski | jaguNET Access Services
> jim@jaguNET.com | http://www.jaguNET.com/
> "Look at me! I'm wearing a cardboard belt!"
>