On Wed, 2008-08-20 at 15:46 +0900, Christian Balzer wrote:
> On the other hand (and I seriously don't want to come across as a fountain
> of negative vibes, but I do realize I probably am) this problem is likely
> to "fix" itself within the next 18 months when I will be forced to support
> IPV6 and thus having to look for some other load balancing solution (and I
> really really like IPVS, I seriously do).
I appreciate that this isn't going to fix things for you right now, but
there is work going on as I type to extend IPVS into IPv6 territory.
There's also some considerable work under way to clean up the existing
code, make it more self-consistent, and fix several possible
races/deadlocks at the same time.
If you can offer some advice, guidance, or coding time then your problem
is more likely to "fix" itself (your quotes) much more quickly.
Unfortunately there's no real timeframe for this, but be assured that it
is happening - check the archives here (and the associated pages):