From: Henrik Nordstrom <henrik@xxxxxxxxxxxxxxxxxxx>
To: azeem ahmad <azeem81@xxxxxxx>
CC: squid-users@xxxxxxxxxxxxxxx
Subject: Re: routing upload requests
Date: Sat, 15 Apr 2006 01:49:45 +0200
fre 2006-04-14 klockan 23:34 +0000 skrev azeem ahmad:
> alright
> but when
> [1-9][0-9]{7}
> mean
> 10000000 to 19999999
> then it means it will match any request more than 10000000 Bytes
yes, as even 54326789012345123 contains at least one of the above
patterns.. (actually several possible matches, 10 to be exact, but one
is sufficient for the acl to be true and regex is satisfied when finding
the first match "54326789")
> then why we need another reg_ex as
> [2-9][0-9]{6}
Because the question was about requests larger than 2 MB (approximately
2000000 bytes), so the acl need to match
2XXXXXX
3XXXXXX
...
9XXXXXX
1XXXXXXX
2XXXXXXX
....
but not
1XXXXXX
as that is below 2 MB.
The first regex matches anything from 10000000 (10 MB) and up.
The second regex takes care of the range 2000000 - 9999999 (2MB to
10MB)
Side note: the second regex also matches many larger values also matched
by the first, but not all. For example it matches 4000000000, but not
10000000. But all these is taken care of with the first regex and does
not matter.
Regards
Henrik
Alright, i got it
mean the two acls are needed because of >2Mb. and if i make it simple to one
mb then i need this
acl large_upload req_header Content-length [1-9][0-9]{6}
cache_peer_access peer2 deny !large_upload
but why it adopted a parser like style, why just a staright definition is
available like in may other options of squid
for example
cache_mem 32mb
Regards
Azeem
