One of the largest consumers of our HTTP bandwidth is Pandora, the
free music service. Unfortunately, Pandora marks its streams as
non-cacheable and also puts question marks in the URLs, which is a
huge waste of bandwidth. How can this be overridden?
--Brett Glass
SS or AUFS?
--Brett Glass
or in town has complained of the same symptom:
weird delays through the cache and none without it.
Is there some popular site out there which has started doing
something that ties Squid in knots?
--Brett Glass
x27;s any way we can help it.)
--Brett Glass
At 09:47 PM 2/28/2009, Amos Jeffries wrote:
>Leaving min at -1, and max at something large (10-50MB?)
>
>Should abort the streams when they reach the max value, You'll have to set the
>max to something reasonably higher than the WU cab size.
>Service Packs may cause issues since they are >100MB
Amos:
We've discovered that Squid won't cache Windows Update unless we set
quick_abort_min to -1. This is because Windows update fetches subranges,
and the subrange fetches individually aren't enough to get the updates
cached. Ideas?
--Brett Glass
At 08:41 PM 2/28/2009, Amos
ble transparent
caching due to problems with streaming?
--Brett Glass
to 5.x, I'm sure.
See
http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/103127
--Brett Glass
one level above standard priority.
Are there "knobs" in squid.conf to control the priorities at which
Squid, diskd, and unlinkd run? Shouldn't there be? And given that
diskd apparently polls for I/O completion on many systems,
shouldn't it have a lower default priority?
--Brett Glass
ng is 2.5STABLE5.
--Brett Glass, LARIAT.NET
The server had already been configured with the line
half_closed_clients off
--Brett Glass
At 02:54 PM 12/8/2005, Matus UHLAR - fantomas wrote:
>On 07.12 23:06, Brett Glass wrote:
>> Just today, I noticed that a transparent Squid proxy running on
>> FreeBSD 4.11 was really b
ion going on between Squid and Akamai.
--Brett Glass
This can tie Squid in knots,
as I've demonstrated in an earlier thread.
--Brett Glass
At 04:36 AM 7/1/2005, Kinkie wrote:
>Huh? Just put your cache_dir in a ramdisk or tmpfs (i.e. /dev/shm on
>recent Linuxes, /tmp on Solaris, don't really know about other OSes..)
If you do that, you lose it when you reboot. And unless it's your
entire cache, you have to split between disk and RAM
At 05:08 AM 7/1/2005, Denis Vlasenko wrote:
>Hmm how much is OS filesystem cache slower? I'd like to use OS fs cache
>for this instead. Otherwise system will have data doubly cached in RAM
>(once in fs cache and once in squid). Than's wasteful.
Most file system caches are limited in size and give
At 04:06 PM 6/30/2005, Henrik Nordstrom wrote:
>I referred to talking to Microsoft here, explaining that also smaller ISPs who
>don't have an Akamai at their core network is in need for some reasonable
>Windows Update distribution method.
I've tried -- both coming in through the front door and
ame URLs.
We could create a "memory disk," but populating it is awkward.
--Brett Glass
;ISPs. But it may simply be an lack of thought.
There's an old saying: "Never assume malice when incompetence will do." ;-)
>Would probably help if some large ISPs tried to talk them to senses, but I
>suppose their preferred solution would be to simply set up an Akamai at the
>ISP..
Akamai refuses to talk to us. Won't even return phone calls. Unless you're a
huge ISP,
they won't give you the time of day. (Which begs the question, "How can you
ever become one
if your bandwidth is all consumed by Windows Update?" The answer, of course, is
that you
must cache on your own.)
--Brett Glass
mall businesses and residential users who do not have
Microsoft servers. Which is not necssarily a bad thing, IMHO, because
Microsoft-based
servers are very unreliable. A small business cannot afford to depend upon
them. Every
small business I know that has tried has had severe problems.
--Brett Glass
ormance, because objects which were previously in RAM
won't be restored to it and will be fetched more slowly thereafter.
--Brett Glass
At 01:20 PM 6/29/2005, Chris Robertson wrote:
>I'm was not trying to make any less of your suggestions, and/or problems.
>My intention was just to provide an alternate solution.
I understand. However, I think that -- given the prevalence of Microsoft
clients and the huge amount of traffic generat
At 11:33 PM 6/28/2005, Odhiambo Washington wrote:
> Looks like someone from Microsoft reads this list for ideas and happened
> to see your long post. They now have addressed your windows update
> problem - files being overly big, by releasing a patch yesterday
I was hopeful when I read this, bu
is to add ACL capability to one Squid parameter?
--Brett Glass
#x27;t call us; we'll call you."
In short, they've put themselves at odds with small ISPs. We would not use
or recommend them.
--Brett Glass
y stored in
the cache. This throws off the statistics, both for byte hit rates and URL hit
rates. All the clients after the first should really be generating "hits,"
because they're being served from disk.
--Brett Glass
tive to do
this, because its updates are currently distributed through Akamai
(which undoubtedly charges it by the bit for downloads). Alas, we
can't hold our breath waiting for Microsoft to do such a thing.
Therefore, the modifications to Squid mentioned above are essential
to providing an efficient solution -- not only to Windows Update
issues but also to issues with similar updating systems from Intuit
and other software vendors.
The first three of these items should be implemented as soon as
possible, so that administrators of Squid caches can safely cache
Microsoft's updates. Now that the largest of these have grown to
more than 700 megabytes, the need is urgent.
--Brett Glass
ith one whose primary drawback is that it could
waste space on the cache disk and/or evict useful objects from the cache.
But it does seem as if it will prevent the huge bandwidth waste we've been
seeing, at least until Squid's behavior is corrected.
--Brett Glass
However,
manual downloads via MSIE are affected too, because the updates
are not cached.
--Brett Glass
is so nasty.
--Brett Glass
ty of your network.
--Brett Glass
At 03:31 AM 4/19/2005, Joel Jaeggli wrote:
>microsoft update makes a pretty serious effort to be uncacheable.
>
>On Mon, 18 Apr 2005, Brett Glass wrote:
>
>>After this month's "Black Tuesday" (the Tuesday on which Microsoft released
resh_pattern download\.microsoft\.com 144000 100% 144000
ignore-reload override-expire override-lastmod
and Squid still reports misses on successive accesses to the same URL.
Can this problem be diagnosed and fixed? It's causing such a
massive waste of bandwidth that we're looking at dumping Squid.
--Brett Glass
ith the source. Would it be possible for someone to add it?
--Brett Glass
files
very well when there's a cache hierarchy. The child cache,
when it gets the request, re-fetches the entire file from the
parent cache in order to cut out the subrange.
--Brett Glass
is not what we're
seeing at all in this case. Windows Update runs through the cache.
And messes it up.
--Brett Glass
f-Unmodified-Since: Sun, 27 Jun 2004 21:36:59
> GMT..Accept-Enc
> oding: identity..Range: bytes=1155018-1157017..User-Agent: Microsoft BITS/6.2..Host:
> download.w
> indowsupdate.com..Via: 1.0 cache1.aabb.net.au:80
> (squid/2.5.STABLE4)..X-Forwarded-For: xxx.2
> 4.234.46..Ca
Squid does NOT handle Windows Update properly. In fact, it multiplies
the traffic by a factor of 1000 in some cases! This is a serious bug
in Squid that really should be fixed.
--Brett Glass
At 03:33 AM 7/15/2004, Lizzy Dizzy wrote:
>Hi,
>
>anyone has their squid cache slowed dow
re-fetching an entire file that's 2 MB or larger from a parent cache.
Needless to say, this is completely killing their bandwidth. Have others
seen this behavior? Is there a known fix?
--Brett Glass
lems, it didn't delete the entries! A
subsequent run revealed all of the same problems.
How can Squid be induced to remove bad entries during the scan of its
database? Preferably without jettisoning the rest of the data?
--Brett Glass
38 matches
Mail list logo
