Chris Robertson wrote:
Elvar wrote:
Elvar wrote:
Hello,
Recently I discovered that one of my clients' bonded T1 links to the
internet was getting maxed out for up to an hour at a time. After
doing some research and using Sawmill to generate some usage
statistics on the user surfing activity, I found that some users
were downloading 1.5 gigs worth of Adobe Reader and other various
downloads. Adobe Reader is obviously not that large and I know the
user would just keep downloading it over and over on purpose. I'm
wondering if somehow Squid is repeatedly downloading these files and
if so, why? I've had squid running at this place for a long time and
I've never seen anything like this. It's a real problem though
because when it happens, the bandwidth is just leveled.
The exact URL which was being accessed the most recent time is
http://ardownload.adobe.com/pub/adobe/re
ader/win/8.x/8.1.3/enu/AdbeRdr813_en_US.msi <javascript:;> .
Is anyone else experiencing something like this lately?
Thanks,
Elvar
Sorry about that, I accidentally forgot to send in plain text. The
URL again is
http://ardownload.adobe.com/pub/adobe/reader/win/8.x/8.1.3/enu/AdbeRdr813_en_US.msi
Could it be that the update is failing and continuously retrying?
Have the following been adjusted from their stock settings:
quick_abort_min
quick_abort_max
range_offset_limit
maximum_object_size
It sounds like the client is requesting the file in sections, but
Squid is grabbing the whole thing. Since it is a 34MB download, it
might be larger than your maximum_object_size (defaults to 4MB), and
as such is not being cached.
Elvar
Chris
This looks very promising. I'm thinking I will probably disable the
aborted fetching period and that will solve the problem. I'll post back
with results after testing it on Monday when the staff is back.
Thanks!
Elvar
