by I'm having a very weird problem with (I think) TCP/IP. I have Linux
installed, up and running nicely on a 486/66 box, 32MB of RAM, and an
NE2000 clone board (ethernet/RJ-45). It works just dandy at the console.
But I am having problems telnetting in from other machines. Oh, it works
- at first. What will happen is that I'll be logged in to the Linux
system, working away happily, when all of a sudden at some point (usually
no more than 10 or 15 minues at most) it will just stop cold. No
communication either way. First I thought it was the network card or
driver, but I switched them (the NE2000 is the replacement) and the same
thing happens. The thing is, the Linux system hadn't crashed - it was up
and running fine if I went to the console to see what was up. The system
I was coming *from* didn't crash, because I could try pinging the Linux
box from other systems and they all reported "no answer."
After a while, I discovered that if I inititiate something from the Linux
console - as simple as a ping to another machine, or a telnet session, or
even a ping to the loopback IP address or the box's own IP address - that
will start things up again. Sometimes it only takes one ping, sometimes
it take a few minutes of random pinging/telnetting. But in short order
things will be fine again and I can go back to my desk, telnet back to
Linux, and be on my way.
Needless to say, this is rather annoying, especially when this Linux
system is supposed to prove to my higher-ups how great Linux is and how
well it works, pending a large investment in Things UNIX. In
desperation, I have told crontab to send a few pings to another machine
once a minute, as sort of a "life support" system to keep the connection
active. I can't tell yet whether this is working or not.
I can't tell if it has anything to do with the *way* I am accessing the
Linux box - I don't *think* so. Generally, I use QVT Net under MS-Windows
with Trumpet Winsock. But other people have experienced the same thing
using other UNIX systems directly for access.
Does anyone have any idea what might be wrong, when Linux's network
services lock up until you go to the console and ping another system, at
which time they work again? I know basically zero about TCP/IP, so I'm
assuming the problem lies theirn somewhere. Probably I just need to
change one silly "packet_maximum_size_foobar_snafu_timing" veriable or
something in some configuration file somewhere. But which one?
Any help GREATLY appreciated! Thanks
----------------------------------------------------------------------
Doug Linder Beware the man of one book.