You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Graham Leggett <mi...@sharp.fm> on 1999/10/15 13:22:40 UTC
Way too many Keepalive requests...?
Hi all,
Recently I have been alerted to the fact that there seem to be many many
keepalive requests that hang around on our server.
This has caused a large number of httpd processes to hang around, many
of them for hundreds of seconds. A netstat confirms that there are
established connections to real clients out there, but these clients are
remaining connected for ages (many many minutes).
Is this normal?
Current Time: Friday, 15-Oct-1999 13:22:06 MET DST
Restart Time: Friday, 15-Oct-1999 13:15:58 MET DST
Parent Server Generation: 9
Server uptime: 6 minutes 8 seconds
Total accesses: 4973 - Total Traffic: 18.2 MB
CPU Usage: u12.08 s9.4 cu8.12 cs15.46 - 12.2% CPU load
13.5 requests/sec - 50.5 kB/second - 3828 B/request
103 requests currently being processed, 11 idle servers
_WK_K_KWK_KKKK_KKWKKKWKK_KKKKKKKKKKKKKKKKKKWKK_KKKKKKKKKKWKKKWW_
KKKKW_KKKKKKKWKKKKKKK_KKKKKKKKWKKWKKWKWWKKW.....K.K.KK...K..._.W
................................................................
................................................................
Regards,
Graham
--
-----------------------------------------
minfrin@sharp.fm "There's a moon
over Bourbon Street
tonight...
Re: Way too many Keepalive requests...?
Posted by Graham Leggett <mi...@sharp.fm>.
Manoj Kasichainula wrote:
> > This has caused a large number of httpd processes to hang around, many
> > of them for hundreds of seconds. A netstat confirms that there are
> > established connections to real clients out there, but these clients are
> > remaining connected for ages (many many minutes).
>
> Do you have a short enough KeepAliveTimeout set?
The default is currently set, 15 seconds. Is this too high?
Some typical lines from /server-status look like this:
50-0 9571 1/21/88 K 1.10 13555 2 1.3 0.14 0.38
131.177.225.136
^^^^^
99-0 26432 9/47/214 K 1.28 8843 2 27.5 0.08 0.53
193.12.220.56
^^^^
A netstat reveals that these IP addresses currently have established
connections to the webserver.
The only thing drastic thing that has changed (to my knowledge) is that
previously the bulk of the content was served by a reverse proxy
configuration, but now it's all served by Apache directly. As soon as
the reverse proxy was taken out the keepalives shot through the roof,
which is killing our webserver and causing much concern.
Any ideas what could be causing this? Broken clients? A broken server?
Is there any workaround we can try? What would be the performance
penalty of switching keepalives off?
Regards,
Graham
--
-----------------------------------------
minfrin@sharp.fm "There's a moon
over Bourbon Street
tonight...
Re: Way too many Keepalive requests...?
Posted by Manoj Kasichainula <ma...@io.com>.
On Fri, Oct 15, 1999 at 01:22:40PM +0200, Graham Leggett wrote:
> This has caused a large number of httpd processes to hang around, many
> of them for hundreds of seconds. A netstat confirms that there are
> established connections to real clients out there, but these clients are
> remaining connected for ages (many many minutes).
Do you have a short enough KeepAliveTimeout set?
--
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/