0

We have a high-availability setup in which distributes requests between a set of backend servers ('realservers' in LVS-speak). LVS is managed by . Each realserver exposes a health check resource over HTTP (basically, if you say GET /info/health to one of them, it will respond with either healthy or ill), and keepalived is configured to poll these resources every 15 seconds to check realserver health. If a realserver becomes unhealthy, keepalived zeroes its weight in the virtual server table, so no traffic is sent to it. All fairly normal stuff, i believe.

Sometimes, it would be really useful to tell keepalived to make a health check for a specific realserver (or all the realservers in a specific service) right now, rather than waiting for the next 15-second point. Is there any way to do this?

For us, this would be useful during deployment operations, where we want to have a new realserver health-checked and added to the pool right away, rather than waiting for it to happen naturally.

Tom Anderson
  • 407
  • 3
  • 12

1 Answers1

1

I think this can be done by sending the HUP signal:

pkill -HUP keepalived
Oliver
  • 5,973
  • 24
  • 33
  • I think that makes keepalived re-read is configuration file. Does that then cause an immediate health check? If it does, it will necessarily be a check of everything, which is not what i was asking for. – Tom Anderson Jun 07 '12 at 16:00
  • Of course `HUP` will do more than you were asking for, but why would rereading the configuration file and checking everything do any harm? Have you tried it? – Oliver Jun 08 '12 at 05:34