Henrik Nordstrom wrote:
On Mon, 21 Mar 2005, Juhani Tali wrote:
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?
Squid will mark the failed address as down and stop using it until the
DNS entry is refreshed (by TTL, client initiated forced reload or PURGE).
Regards
Henrik
I tested the behaviour a bit more:
squid.conf
only one cache_peer line:
cache_peer cache_test.aso.ee parent 3128 7 no-query round-robin
and tcpdump:
09:50:37.971722 IP kontor.aso.ee.37646 > cache1.aso.ee.3128: . ack 265
09:50:38.008114 IP kontor.aso.ee.37646 > cache1.aso.ee.3128: . ack 1539
09:50:38.072213 IP kontor.aso.ee.37855 > 195.80.96.217.3128: S
09:50:38.072383 IP kontor.aso.ee.37855 > 195.80.96.217.3128: . ack
09:50:38.072755 IP kontor.aso.ee.37855 > 195.80.96.217.3128: P
09:50:38.202032 IP kontor.aso.ee.37646 > cache1.aso.ee.3128: . ack 2959
09:50:38.207653 IP kontor.aso.ee.37646 > cache1.aso.ee.3128: . ack 3149
Now without the round-robin:
cache_peer cache_proov.asosise parent 3128 7 no-query default
09:59:33.029014 IP kontor.aso.ee.37870 > cache1.aso.ee.3128: . ack
09:59:33.029078 IP kontor.aso.ee.37869 > 195.80.96.217.3128: P
09:59:33.029166 IP kontor.aso.ee.37870 > cache1.aso.ee.3128: P
09:59:33.204759 IP kontor.aso.ee.37870 > cache1.aso.ee.3128: . ack 351
09:59:44.679716 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 265
09:59:44.695619 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 1539
09:59:44.896929 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 2959
09:59:44.899308 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 3149
09:59:44.918317 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 4569
09:59:44.919612 IP kontor.aso.ee.37870 > cache1.aso.ee.3128: P
09:59:44.933504 IP kontor.aso.ee.37869 > 195.80.96.217.3128: . ack 5855
