[dnsdist] server policy to failover to another pool

[Remi Gacogne]

Hi Justin,

On 10/06/2017 02:06 PM, Justin Valentini wrote:
> I tried the orderedLeastOutstanding policy but ran into issues when
> testing. If I explicitly set my order #1 server to down (e.g.
> getServer(0):setDown()) dnsdist doesn't not directly traffic to the next
> ordered server. Instead the query times out. However, when we bring down
> the actual dns server and let dnsdist healthcheck it to down, it seems
> to work.
> 
> Do any of you know why this may be the case? Do isUp() and upStatus not
> respect explicitly set DOWN state?

The way it is currently handled might not be very intuitive. When you
call setDown() or SetUp(), this sets the backend's "availability" state
to Up or Down, instead of the default of Auto, but does not change the
"upStatus".
The isUp() method checks whether "availability" is set to Down or Up
before checking "upStatus, so orderedLeastOutstanding should either use
the isUp() method, or check the "availability" state before looking at
"upStatus".

-- 
Remi Gacogne
PowerDNS.COM BV - https://www.powerdns.com/

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.powerdns.com/pipermail/dnsdist/attachments/20171006/5b4f13cf/attachment.sig>