Re: keepalived may broke ipvs one-packet scheduling

Maybe I'm wrong, by explanation of Julian Anastasov, I think it's all
because keepalived's not supporting of --ops in ipvsadm.

2013/8/26 Drunkard Zhang <[email protected]>:
> If start keepalived before apply --ops of ipvsadm, the whole ipvs
> subsystem seems broken, here it is:
>
> 1. --ops not working, but the config in /proc/net/ip_vs all looks reasonable.
>
> 2. Conns and Conns/s in /proc/net/ip_vs_stats_percpu are always a
> fixed little number, but it's a counter, should always changing by
> traffic.
>
> 3. .ipvsadm -C doesn't fix it, but ./ipvsadm -C && rmmod ip_vs_wrr
> ip_vs && ./ipvsadm -R rules-with-ops" works
>
> Here's full story:
> http://archive.linuxvirtualserver.org/html/lvs-devel/2013-08/msg00064.html
>
> Here's extra information:
> I'm using keepalived-1.2.7 and git version of ipvsadm while tesing,
> and I'm using wrr as scheduling  algorithm.

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk