Article delegate-en/695 of [1-5169] on the server localhost:119
  upper oldest olders older1 this newer1 newers latest
search
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]

Newsgroups: mail-lists.delegate-en

[DeleGate-En] Delegate 6.0.9alpha on WinNT/2K problems/bugs/suggestion
13 Feb 2000 11:59:19 GMT "Matej Mihelic" <pwubabdyi-k2yktwbwo5br.ml@ml.delegate.org>


Hello.

I have tried to replace our current proxy with Delegate and I stumbled over
some roadblocks. Here they are:

*** NNTP specialist problem (in versions 6.0.9 and 5.9.13):

If I try to request a large number ob articles over a slow link I'll always
receive "Server: 503 Timeout after 300 seconds, closing connection..". This
happens in the middle of a working session (while Delegate is reading from
server or sending data to client)

I have - at least I hope so - found the problem in the following line:

nntp.c:4995 int CC_TIMEOUT_NNTP = 300;

I have found no way to override this built in value. TIMEOUT=nntpcc:9000 has
no apparent affect.

*** Severe NT performance problem while increasing number of concurrent
connection to any kind of Delegate server/proxy process. (version 6.0.9)

Delegate seems to find "ghost" problems that do not really happen and tries
to retry the operation after a random and increasing delay. The performance
is improved a lot by setting all delays to 0 (DELAY=unknown:0,
DELAY=reject:0). I have no idea why but I was able to reproduce it on three
systems with WinNT4SP6a, Win2K and Win98SE2.

*** Two enhancement suggestions for CFI filters:

1) I have a CFI filter, that should process all text/html files smaller (or
larger) than 10000bytes. It would greatly increase the performance, if I
could specify this limit in the CFI definition instead of checking in the
CGI program. Perhaps you could add two more rules to the CFI parser:

"Content-Length-Greater-Than:"
and
"Content-Length-Smaller-Than:"

User should be able to specify only one or both values (so one could check
for >,< and between).

2) It would be easy to create an efficient a "JunkBuster" CFI filter if I
could specify search rule for the entire requested URL. With "Req-Url" I can
scan the "path" part, but I have found no way (beside a CGI program) to scan
the "host name" (I was testing on a HTTP/PROXY). I would really like to find
all hosts with names containing "ad.", "ads.", ".images." and similar.

Perhaps some kind of generic search over header fields - as simple list of
word that must appear on the same line (AND) - would be the best solution:

Content-Type: text/plain
Header-Find: MIME 1.0
Header-Find-Not: Encoding gzip
Header-Find-Not: Encoding deflate

Would match all text/html documents that contain header line with words
"MIME" and "1.0" and do not contain header line with "Encoding" and "gzip"
or "Encoding" and "deflated".


  admin search upper oldest olders older1 this newer1 newers latest
[Top/Up] [oldest] - [Older+chunk] - [Newer+chunk] - [newest + Check]
@_@V