[squid-users] time based range_offset_limit

[Heiler Bemerguy]

10.1.10.9 is the proxy's IP, dude

And it is connecting to the same server multiple times because a client 
is doing a RANGE download.. (windows update to be exact)

All GETS are like these:

GET 
http://au.v4.download.windowsupdate.com/d/msdownload/update/software/defu/2016/06/am_base_7684a3445029744f69529e465ac573f76bd68144.exe 
HTTP/1.1
Accept: */*
Accept-Encoding: identity
If-Unmodified-Since: Wed, 29 Jun 2016 22:14:50 GMT
*Range: bytes=103416560-104234280*
User-Agent: Microsoft BITS/7.8
Proxy-Connection: Keep-Alive
Host: au.v4.download.windowsupdate.com

But with an increasing *range *every ~2 secs.... and squid will create a 
new connection to the server everytime

Anyway.. I'm trying to figure out this by myself for months now..

In my "test lab" it seems collapsed_forwarding and range-requests are 
working together.. but in this production server I always get this 
behaviour.. lots of connections to the same server, and to get the same 
file..

How to get rid of it without giving up on caching it? lol


-- 
Best Regards,

Heiler Bemerguy
Network Manager - CINBESA
55 91 98151-4894/3184-1751

Em 12/07/2016 17:08, joe escreveu:
>>> root at proxy:~# netstat -n |grep 201.30.251.27 |grep ESTAB
>>> tcp   243802      0 10.1.10.9:27788 201.30.251.27:80        ESTABELECIDA
>>> tcp        0      0 10.1.10.9:15343 201.30.251.27:80        ESTABELECIDA
>>> tcp    14480      0 10.1.10.9:32548 201.30.251.27:80        ESTABELECIDA
>>> tcp        0      0 10.1.10.9:25426 201.30.251.27:80        ESTABELECIDA
>>> tcp    48322      0 10.1.10.9:8560 201.30.251.27:80        ESTABELECIDA
>>> tcp   329234      0 10.1.10.9:54205 201.30.251.27:80        ESTABELECIDA
>>> tcp        0      0 10.1.10.9:1656 201.30.251.27:80        ESTABELECIDA
>>> tcp      993      0 10.1.10.9:50820 201.30.251.27:80        ESTABELECIDA
>>> tcp   330227      0 10.1.10.9:56519 201.30.251.27:80        ESTABELECIDA
> 10.1.10.9  one client ip i cant tell he might be downloading more then one
> file
> range_offset_limit -1 or none it will force range to non range download
>
>>> Look my conf:
>>> acl fullDLext urlpath_regex -i
>>> \(exe|ms(i|u|p)|deb|cab|rpm|bin|zip|ax|r(a|p)m|app|pkg|mar|nzp|dat|iop|xpi|dmg|dds|thor|nar|gpf|pdf|appx|appxbundle|esd)
> ouchhh 2 much unless you have plenty of bandwith
>
>>> acl fullDLurl url_regex -i \.microsoft\.com\/filestreamingservice
>>> quick_abort_min 0 KB
>>> quick_abort_max 0 KB
>>> quick_abort_pct 100
>>> range_offset_limit -1 fullDLext
>>> range_offset_limit -1 fullDLurl
> better to change that to   one control i dont know if that will be bad idea
>
>
>>> collapsed_forwarding on
> this is good if you have other client downloading on same time same file it
> has nothing to do with multi connection to the same  ip only it will help
> saving you example if you have 10 client downloading at the same time same
> file so one download it will be.. not 10 download i dont know if im
> mistaking amos or other will answer to that
>
>
>
>
>
> --
> View this message in context: http://squid-web-proxy-cache.1019090.n4.nabble.com/time-based-range-offset-limit-tp4678462p4678468.html
> Sent from the Squid - Users mailing list archive at Nabble.com.
> _______________________________________________
> squid-users mailing list
> squid-users at lists.squid-cache.org
> http://lists.squid-cache.org/listinfo/squid-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squid-cache.org/pipermail/squid-users/attachments/20160712/3dbcc47a/attachment.html>