lkml.org 
[lkml]   [1999]   [Mar]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
Subject2.2.3-ac1: route or 3c509 bug?
Alan, Alexey, et al,

I encountered an odd networking bug with 2.2.3-ac1 (cannot reproduct,
*sigh*, I know, I know), so this report may only be useful if somebody else
reports it too; it could be in 2.2.3. I had a box start delaying its
packets by about 20 seconds for no obvious reason. I was about to ^C my
ping command when the echoes starting coming back. I was at home, the box
at work, so I had little I could do to effectively diagnose remotely.

Two somewhat antiquated boxes (P75 UP, 3c509, 16Meg, 2x 1G IDE) act as a
high-availability 24x7 cluster in the hospital environment here; the backup
makes it an exact duplicate of the master every 20 minutes using rsync. Two
weeks ago, I upgraded the slave from 2.0.36 to 2.2.2, left the master
running at 2.0.36. No problems in two weeks, so I decided over the weekend
to upgrade both to 2.2.3-ac1. The master (with all the user load) has been
up and doing wonderfully. But within 18 hours, something happened to the
slave's networking.

Noting that packets were making it back OK, just 20 seconds delayed, I
telnetted to the SYSTAT port (== "ps aux") to see what processes were
running. One tea-time later, I saw a very normal (and rather quiet) looking
output from `ps`. Tcpdump, OTOH, showed a box that was regularly sending
out "arp: who has a.b.c.d tell me" followed immediately by the router
identifying itself, or the master box identifying itself, as the case may
be. A variety of ntp packets went out, and servers responded, but I have
little clue what the ntpd process itself saw for packet latency (that would
have been interesting).

Sadly, in my haste to get the slave working again, I neglected to do a
ifconfig;route;route-C via rsh before I did a reboot. Upon reboot, the logs
were devoid of any warnings or clues whatsoever. No failed route messages,
no 3c509 errors, not even bogus clock messages from ntpd.

The only non-vanilla config for the box is that IP firewalling and aliasing
are both enabled, though the ipchains rules are currently empty. (The box
was root-compromised and trojanised from somebody at telepac.pt who istalled
a sniffer called, ironically, "(kswapd)", so I wanted to shore the box up at
some point.)
The primary eth0 has static IP's, one for box1, one for box2; eth0:0 has
changeable IPs depending upon the role of the box, either master or slave.
Ifconfig and route are from net-tools-1.50.

Cheers,
Kris

--
Kristofer Karas * kXtXk@bigfoot.com (delete X's)
AMA/CCS DoD RF900RR NT650++ !car * Senior systems engineer/SysAdmin
"Build a system that even a fool can use, * BI Deaconess Medical Center, Boston
and only a fool will want to use it." * Will design LISP machines for food




-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:50    [W:0.099 / U:0.064 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site