On Thu, 28 Jun 2001, Charles Cazabon wrote:
>Err -- this is the whole problem. If you're running inetd, I can get your
xinetd != inetd. Have you tested and seen that this is still a problem?
>system to shut off the POP3 service for ten minutes simply by initiating five
>or ten connections in a few seconds. This is a bug, not a feature.
Yes, I know, very well, and this is the reason I would never put anything
like a SMTP server behind inetd.
>On the contrary; tcpserver lets you set concurrency limits, which inetd cannot
>do.
Concurrency-limits per connection from IP-adresses? How? Please tell me.
As far as I can see, you need to supply tcpserver with the normal
cdb-file, and yes, you can give different enviroment variables per
ip-range, but this file is static. Can you do more dynamic resource
management? As far as I can see, no (and I don't regard regenerating a cdb
file every half a minute as a good solution).
And, no, I don't know if xinetd is good as this eiter, that is what I am
planning to test (on a large scale system)
(But, don't misunderstand me, I don't have it in for tcpserver in any way,
I like the program and are using it several places. I just reacted to your
way of dismissing everything else).
--
Thorkild
