[ Butchering the reply because of stupidity and webmail ] on July 22, 2019, at 10:03:18 PM, "Ask Bjørn Hansen" <[email protected]> wrote: > [ resending from an account that doesn’t have -all in it’s spf record ]
>> If there was a way to do that, it would be easy to update servers to check >> occasionally and stop using servers that have left the pool. > For users using the “pool” keyword, I think it’d make sense to every ~24 > hours (?) replace the worst performing server and, for example, every week > (?) replace the “oldest” server. > The reason to replace the oldest server (if it’s still in the pool or not) > would be to continuously load balance between all the available servers, > rather than have clients hold on to servers that then get a slowly increasing > load. If “pool” configured clients were the majority and they just had a > policy of always (slowly) rotating servers, this shouldn’t happen. Just think of it, you could completely remove a servers' traffic in months (dare we hope). Instead of the current which is like "My grandparent removed the server from the pool in their youth the pumpkins still send tsunamis and I am old." >> On Jul 21, 2019, at 14:51, Hal Murray <[email protected]> wrote: >> That won't catch the case where the pool DNS name is used as a server. If we >> put a simple still-there check in the server case, clients would end up >> switching servers every time they checked. I don't see any harm in that. > I don’t think I understood what this meant. There are pumpkins have the pool added via multiple server directives. He does not see any harm in a check to only replace dysfunctional/absent servers at config (re)load. I'm sure I misread that. _______________________________________________ pool mailing list [email protected] http://lists.ntp.org/listinfo/pool
