[squid-users] RANGED requests still a pain for me..

[Amos Jeffries]

On 26/10/2016 2:26 a.m., Heiler Bemerguy wrote:
> These firefox updates takes weeks to be really cached.. it won't cache
> if more than 1 person is trying to download at the same time.. but WHY
> does this happen?
> I've filled a bug report about this *TCP_SWAPFAIL_MISS*, but it seems no
> one is interested.. lol RockStore BTW
> 
> *acl fullDLext urlpath_regex -i
> \.(exe|ms[iup]|cab|bin|zip|mar|pdf|appx(bundle)?|esd|lzma2)\??**
> **range_offset_limit -1 fullDLext*
> 
> 1477401191.608   5151 10.61.0.67 TCP_MISS/206 300542 GET
> http://download.cdn.mozilla.net/pub/firefox/releases/48.0.2/update/win32/pt-BR/firefox-48.0.2.complete.mar
> - HIER_DIRECT/200.149.150.209 application/octet-stream
> 1477401192.592 206375 10.20.1.2 TCP_MISS/206 300544 GET
> http://download.cdn.mozilla.net/pub/firefox/releases/48.0.2/update/win32/pt-BR/firefox-48.0.2.complete.mar
> - HIER_DIRECT/200.149.150.203 application/octet-stream

The log does not tell us much except that it was a cache MISS and that a
206 was delivered to the client.

Also, note that the duration column matters a lot. It indicates that the
second of those entries was started and still ntot yet completed when
the first happened.


A "debug_options 11,2" cache.log trace with the full HTTP message
headers to/from client and to/from server for these is neeed to see what
exactly is going on at the traffic level.

A deeper (ALL,5) cache.log trace might be needed to see what is causing
that traffic level messaging to happen if its not obvious from the 11,2
details.

Amos