The DNS round robin should work (and I am using it at the moment, somewhat).
The more interesting is what will happen if one server goes down. Will the child squid take the next address from dns or fail?
Tests showed that it will take the next address and it will not fail.
Now I have another problem, how can I force the clients to rebalance after the second server comes up again?
At the moment each of the 3 squid processes is serving around 160 req/s, so after one server down and up again some rebalance is quite needed ;)
I do not have access to the 1-st level squid-s.
Juhani
nikolay.nenchev@xxxxxxxxxxxxxxxxxxxxxxx wrote:
With DNS you making round robin, and in this case every request is send to the next in the zone file machine. like that 1,2,1,2,1,2 and because squid is loading and checking dns at startup is it going to use this round-robin. But be sure that you should restart squid after modifing zone files in named and named restart.
Nikolay
######################################################################
Hi
I have a 2 level setup
1-st level is local squid, eg. squid-office that forward all non-local traffic to 2-nd level squids.
2-nd level are 4 parent squid servers: squid1, squid2, squid3, squid4. (2 servers, 2 processes to use both cpu-s)
It would be nice to load-balance the parents and if one of the parents should go down then to make it more reliable.
---
One way might be to add to all 1-st level squids parents to squid-office.conf
cache_peer cache1 parent 3128 7 no-query no-digest default weight=2
cache_peer cache2 parent 3128 7 no-query no-digest default weight=1
....
Will this work with no-query?
---- Will the following do load-balancing and failover? (I saw something like this from Henrik-s homepage) DNS: cache A 192.168.2.1 #cache1 cache A 192.168.2.2 #cache2 and then to squid.conf cache_peer cache parent 3128 7 no-query no-digest default weight=1
Juhani
