502 errors with apache mod_proxy hot standby (or equivalent)
Solution 1:
If you explicitly disable the primary balancer member first (before bringing down the primary server), Apache will happily route request to the secondary balancer member. Takeover happens immediately.
To do that, enable the balancer manager first with config:
<Location /lb>
SetHandler balancer-manager
</Location>
Then, you can interactively or programmatically enable / disable balancer member by accessing /lb.
Contrary to karmawhore's comment, a hot-standby setup like this has nothing to do with health-check.