Hello,
Sorry about my wrong answer earlier -- thanks for the correction.
The short version of my problem is that up to squid 2.5 stable 4, I
seemed to
get automagic retrying of requests on alternative parent caches, but
when I upgraded to 2.5stable10 and then 2.6stable3 I lost that
behavior. Can I change the configuration to get that behavior back
with the
newer versions?
The reason I ask is that I have a somewhat unusual reverse proxy setup
where I have front squid caches relaying requests to a virtual domain
that contains squid proxies which relay requests to the compute code.
I.E. the compute servers are parent caches in a particular domain, and
each compute server has a squid/compute demon pair.
The main reason it is set up this way is that up-to-and-including squid
2.5stable4, the front caches would automatically retry failed (GET)
requests on another cache. So to upgrade compute software I could
simply take out the compute demon while keeping the local squid running,
and the requests would get rerouted.
This was true even when I upgraded the compute caches to squid
2.5stable10.
However, when I upgraded the front caches to 2.5stable10, I started
seeing errors instead, e.g.
1161025036.587 203 has-224.dhcp.umass.edu TCP_MISS/503 1544 GET
/favicon.ico - FIRST_PARENT_MISS/iridlc6p.ldeo.columbia.edu text/html
[Host: ingrid.ldeo.columbia.edu\r\nUser-Agent: Mozilla/5.0 (Windows; U;
Windows NT 5.1; en-US; rv:1.8.0.7) Gecko/20060909
Firefox/1.5.0.7\r\nAccept: image/png,*/*;q=0.5\r\nAccept-Language:
en-us,en;q=0.5\r\nAccept-Encoding: gzip,deflate\r\nAccept-Charset:
ISO-8859-1,utf-8;q=0.7,*;q=0.7\r\nKeep-Alive: 300\r\nConnection:
Close\r\n] [HTTP/1.0 503 Service Unavailable\r\nServer:
squid/2.5.STABLE10\r\nMime-Version: 1.0\r\nDate: Mon, 16 Oct 2006
18:57:16 GMT\r\nContent-Type: text/html\r\nContent-Length:
1073\r\nExpires: Mon, 16 Oct 2006 18:57:16 GMT\r\nX-Squid-Error:
ERR_CONNECT_FAIL 111\r\nX-Cache: MISS from
iridlc6p.ldeo.columbia.edu\r\nProxy-Connection: close\r\n\r]
Since I just changed the front cache, presumably 2.5stable4 was
handling this error by retrying the request on another cache.
So can I get the retry behavior of the front caches back?
Alternative question, is there a way to signal the squid running on
the compute server to refuse requests while completing any current
requests? Some of these requests can be slow, so it might be in this
state for tens of minutes.... I suppose I could change the config file to
refuse ICP and reread it, but that seems pretty clunky....
Thanks,
Benno
--
Dr. M. Benno Blumenthal benno@xxxxxxxxxxxxxxxx
International Research Institute for climate and society
The Earth Institute at Columbia University
Lamont Campus, Palisades NY 10964-8000 (845) 680-4450
