Date: Tue, 7 Apr 1998 09:17:47 -0600 (MDT) From: R Ricci <R.Ricci@m.cc.utah.edu> To: freebsd-isp@FreeBSD.ORG Subject: Re: VERY strange network(?) problem Message-ID: <Pine.GSO.3.96.980407090746.25596A-100000@bobo> In-Reply-To: <199804070753.AAA07571@implode.root.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Looking through the messages files, there are no messags about mbuf cluseters. This kernel was not compiled with NMBCLUSTERS, and maxusers is set at 50. Our webserver is relatively low-volume, serving only a few thousand hits/day, and supplying mail and DNS for about two hundred people. I would happily recompile the kernel (in fact, I'm doing so right now ith 75 maxusers) it it will solve this problem. Here's another tidbit to add to the confusion: I can telnet in and `cat /var/log/maillog`, which is pretty long, without a problem, but if I try to cat my mail file, it stalls after about a dozen lines. Even if I move this mail file into the /var/log directory, so that it is on the same drive as the mailllog, it still stalls in almost exactly the same place. Robert Ricci rricci@theonlynet.com On Tue, 7 Apr 1998, David Greenman wrote: > >I'm having a very strange problem with our web/mail server (running > >2.2.1), and I'd appreciate any pointers any of you could give me. Part of > >my problem is that I'm not sure exaclty what methods I can use for > >tracking down the source of my difficulties. > > > >Often, during TCP connections to the server, the connection will "stall", > >ie. I get no data for several seconds/minutes. This causes huge problems, > >of course, when it comes to checking mail, retrieving web pages, etc. At > >frist, I suspected disk problems, but experimentation has pretty much > >ruled this out. Doing a 'ps' on the server while one of these stalls is > >occuring reveals that the offending process is just in state S or I, not > >waiting on the disk. > > > >The only clue I have (and it may not be relevant) is the following > >message, which keeps showing up in /var/log/messages/ : > > > >arplookup 207.201.65.2 failed: host is not on local network > > > >This IP is on our gateway. It's another interface on that gateway, but not > >on our local network. (Our server sees the gateway as 207.201.125.65). I'm > >not sure what I need to do do fix this. > > > >Any help, and/or suggestions of additional things to try will be very > >welcome! > > Sounds like you might be running out of mbuf clusters. Have you specified > an amount with the NMBCLUSTERS kernel option? If not, what do you have > maxusers set to in your config file? Also, about how many simultaneous > TCP connections do you expect have at any time? A rough number of http > hits/day might also be useful in calculating an appropriate number of > buffers. > To confirm this is the cause, you might also look for the message "Out of > mbuf clusters" in your /var/log/messages file. > > -DG > > David Greenman > Core-team/Principal Architect, The FreeBSD Project > To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.GSO.3.96.980407090746.25596A-100000>