[2649] in linux-net channel archive
Linux 1.4/2.0
daemon@ATHENA.MIT.EDU (Bernhard Heidegger)
Thu Apr 25 11:42:48 1996
Date: Mon, 22 Apr 1996 19:48:08 +0200 (MET DST)
From: Bernhard Heidegger <hdg@edvz.tu-graz.ac.at>
To: linux-net@vger.rutgers.edu
Hi!
Is it possible to raise the (static) max. open filedescriptors per
process before 1.4/2.0?
IMHO, 256 fd's is to less today; think of (new) WWW servers which
supply connection oriented (keepalive) HTTP. I think, 256 connections
is to less for a middle WWW server.
I've tried to modify (IMHO) all #defines and recompiled the kernel
(something between 1.3.55 and 1.3.65); but the machine didn't boot anymore.
I had interesting problems: when I booted to single user mode all went
well, and than I could start network services (also portmap, nfsd,
mountd, and a simple testserver; I think syslogd didn't work). But from
time to time the machine crashes...
And now a "real" problem:
We have a Pentium-166 with 64MB, AHA2940UW, 2 Quantum XP34300W 4GB and
a 3c509. This machine runs 1.3.74 (ELF based system) with the following
increased values in linux/include/linux/fs.h:
#define NR_INODE 4096 /* this should be bigger than NR_FILE */
#define NR_FILE 4096 /* this can well be larger on a larger system */
We run our Hyper-G server (a better WWW server) which uses a type of
keepalive sessions on this machine, but from time to time the kernel prints
tcp_send_skb: attempt to queue a bogon.
in a loop on the console (this happened 4 times in about 2 weeks). I think
the kernel didn't schedule anything after that; the key combinations like
ALT+SCROLL seems to work, but I am unable to see the results because of
the fast tcp_send_skb... prints.
I will install >= 1.3.95 in the next days, but I think this bogon check
didn't change since 1.3.74?
Thanks for any suggestion
Bernhard.
PS: I will mail any experience with the new kernel...
---
+----------------------------+-------------------------------+
| hdg@edvz.tu-graz.ac.at | bheide@iicm.tu-graz.ac.at |
+----------------------------+-------------------------------+
| Bernhard Heidegger, Graz University of Technology, Austria |
+------------------------------------------------------------+
Worst day playing is better than best day working!
