[lvs-users] How ldirectord polls real servers

Malcolm Turnbull malcolm at loadbalancer.org
Tue Apr 23 17:26:08 BST 2013


Gugliemo,

Ldirectord has various different health checking methods.

For a simple TCP based service you could use the 'connect' check i.e.
Can I do a TCP open on that port?
For UDP you would probably just use a ping check as it is connectionless.

The various methods are pretty well documented in the man page.



On 23 April 2013 16:42, Guglielmo Abbruzzese <g.abbruzzese at resi.it> wrote:
> Hi,
> as far as you know where can I find a detailed  description/documentation
> about how ldirectord polls the realservers in order to realize if the node
> should be actually removed from the cluster?
>
> For sure, " ldirectord monitors the health of the real servers by
> periodically requesting a known URL and checking that the response contains
> an expected response. If a real server fails then the server is removed and
> will be reactivated once it comes back on line.".
>
> But what if the real servers implement a custom tcp or udp service?
> In case of a tcp persistent connection, what if there is no data flow for a
> while (shall I rely just on the 'persistent' timeout)?
> And what about a cistom udp server?
>
> Is there a way to use tcpdump in order to detect such evidence?
>
> Thanks in advance for your advice,
> Regards
> G.
>
>
>
> _______________________________________________
> Please read the documentation before posting - it's available at:
> http://www.linuxvirtualserver.org/
>
> LinuxVirtualServer.org mailing list - lvs-users at LinuxVirtualServer.org
> Send requests to lvs-users-request at LinuxVirtualServer.org
> or go to http://lists.graemef.net/mailman/listinfo/lvs-users



-- 
Regards,

Malcolm Turnbull.

Loadbalancer.org Ltd.
Phone: +44 (0)870 443 8779
http://www.loadbalancer.org/



More information about the lvs-users mailing list