|
Sorry guys, I kind of dropped off the face of the earth for a couple of
days. I applied the patch and am still having the same problem. I Have
attached so debug output (gziped). Below is an updated rules file (same
hardware, slightly different ips).
#!/bin/sh
/sbin/ipvsadm -C
/sbin/ipvsadm -A -t 172.19.0.6:25 -s lc
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.3 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.4 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.5 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.6 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.7 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.8 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.9 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.11 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.12 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.13 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.14 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.15 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.17 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.18 -m
/sbin/ipvsadm -a -t 172.19.0.6:25 -R 172.18.2.23 -m
172.18.2.4, 172.18.2.7, 172.18.2.8 were still recieving email when I
took the debug output. The rest were not
Thanks,
Jason
On Sun, Apr 01, 2001 at 10:58:14PM +0800, Wensong Zhang wrote:
>
>
> Hi,
>
> Please apply the attached patch to ipvs-0.2.7
> cd <your_directory>/ipvs-0.2.7/ipvs
> patch -p0 < <your_directory>/ipvs-027-1.diff
>
> Then, rebuild the ipvs modules and see if the same bug happens again.
>
> The attached includes two fixes:
> * Fixed the wrong setting expiration of ipvs connection entries
> which would lead to the wrong add_sltimer/del_sltimer calls in the
> SMP environment.
>
> * Julian fixed the wrong ip_vs_conn_put in icmp handling code.
>
> Thanks,
>
> Wensong
>
>
> On Tue, 27 Mar 2001, Jason Steenblik wrote:
>
> > Problem:
> > After about 3 or 4 hours of steady traffic through the load balancer
> > using least connection the real servers begin to be excluded until only
> > one is recieving traffic. I have 16 real servers that all recieve
> > traffic initally but one by one they go quiet until eventually all
> > traffic is sent to just one server. On the other hand, round robin
> > works fine through weeks of continual traffic. My logs are filled with
> > thousands of the following message with both scheduling methods:
> >
> > kernel: bug: kernel sltimer added twice at f88a541e.
> >
> > Box info:
> > Kernel: 2.4.2
> > ipvs: 0.2.7, 0.2.6, & 0.2.5
> > cpu: 2 x 860Mhz
> > Ram: 1GB
> > Distro: RedHat 7.0
> >
> > Modules Loaded:
> > Module Size Used by
> > ip_vs_lc 1008 1 (autoclean)
> > ip_vs 552961876 (autoclean) [ip_vs_lc]
> > eepro100 17232 2 (autoclean)
messages.gz
Description: application/gunzip
|
