Date: Wed, 10 Nov 1999 22:06:58 -0800 (PST) From: "Rodney W. Grimes" <freebsd@gndrsh.dnsmgr.net> To: stabilizer@klentaq.com (Wayne M Barnes) Cc: billm@danger.ms (Bill Marquette), freebsd-stable@FreeBSD.ORG (FreeBSD Stable) Subject: Re: ethernet hard or soft failure Message-ID: <199911110606.WAA79532@gndrsh.dnsmgr.net> In-Reply-To: <199911110514.XAA01984@klentaq.com> from Wayne M Barnes at "Nov 10, 1999 11:14:51 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
> Dear Bill > > :( > > Our network guru has looked for rogue cards or laptops on our > dept. network and found none. Has he put a thermometer inside your computer case.... that would be my next step. Also for the fellow below with the 21x4x cards, watch out there have been a lot of bad chips since Intel took the production of this over, they tend to stop talking on the wire anywhere from 15 min to 2 days of power on. Seems to be thermally related. Kingston now replaces them without even a question when I call with another one doing this. The problem is not resticted to Kingston cards as I have seen it on generic cards as well, and seems to inflict only the DC21x4x chips with the word ``Intel'' on them, the ones that say ``Digital'' have always worked just fine :-) > I have increased maxusers to 256, and have tried NMBCLUSTERS at 1024, > 4096, (and currently 16K; too recent for results, yet.) > > I have increased my bpf units to 24. > > I have set up a cron job to run every 15 minutes and log the > DIFF against the last time, then DATE, to > http://barnes1.wustl.edu/wayne/NMB > [Here is testNMB* which crontab runs every 15 minutes: > netstat -m | diff NMB.boring - | grep ">" >> ~wayne/NMB.log > date >> ~wayne/NMB.log > netstat -m > ~wayne/NMB.boring > ] > > No luck so far. Every weekday between 1 and 4 or so, my NIC card stops > transmitting across the network. The console is stays up, but no pings > can get out, nor can anyone log in from anywhere else. What does the link light on the hub show is going on? Usually when I see this thermal problem the hub light will either be going nuts, or on the partitioning hubs it shows the port has gone into jabber mode and has been partitioned out of the network. > Our network guy JJ found a recommendation on "Deja News" (I don't know > where that is) to increase kern.vm.kmem.size as the next resort. I > don't know what a the current value is, or what a reasonable next level > is. I can't tell by inspecting source mbuf.h, either. It's well > concealed, if it's there. > > Wayne M Barnes stabilizer@klentaq.com > > > Wayne, I'm curious to see if you ever got your ethernet issues resolved. > > I'm currently having VERY similar issues with a -CURRENT box. My NIC's > > have been 3com 3c905 and generic DEC 2104x chipset cards, FYI. So far for > > me the only thing that's helped (hasn't fixed it, but helped) has been to > > disable my screensaver in X. This could perhaps lower the thermal temperature in your chassis, please check your ventalation and internal temperature. Look for dead power supply fans, etc. Oh, and look for an Intel branded DC21x4x chip... -- Rod Grimes - KD7CAX @ CN85sl - (RWG25) rgrimes@gndrsh.dnsmgr.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199911110606.WAA79532>