Date: Thu, 5 Dec 1996 17:21:07 +0100 (MET) From: Greg Lehey <grog@lemis.de> To: pabraham@PAbraham-S.mankato.msus.edu (Paul Abraham Mullaseril) Cc: questions@FreeBSD.org (FreeBSD Questions) Subject: Re: Buffer Message-ID: <199612051621.RAA25361@freebie.lemis.de> In-Reply-To: <Pine.BSF.3.91.961205153258.1027C-100000@PAbraham-S.mankato.msus.edu> from Paul Abraham Mullaseril at "Dec 5, 96 03:46:00 pm"
next in thread | previous in thread | raw e-mail | index | archive | help
Paul Abraham Mullaseril writes: > Hi Greg, > My apologies for not being specific. The version of BSD is 2.1.5 (August > 1996). I notice the problem occurs wheneever there is an external access > to the machine via ftp or web server (apache) or I send a mail with an > attachment like a large postscript file etc. The machine has a 3com > Etherlink 111 network card (specifically 3C590-TPO). I have collected > some statistics that may be usefull. > > The output from netstat -m is > 145 mbufs in use: > 69 mbufs allocated to data > 16 mbufs allocated to packet headers > 44 mbufs allocated to protocol control blocks > 16 mbufs allocated to socket names and addresses > 5/48 mbuf clusters in use > 114 Kbytes allocated to network (24% in use) > 0 requests for memory denied > 0 requests for memory delayed > 0 calls to protocol drain routines So far, this looks fine. Plenty of buffers available. > The output from ping 134.29.1.1 (name server) is: > ping: sendto: No buffer space available > ping: sendto: No buffer space available > ping: sendto: No buffer space available > PING 134.29.1.1 (134.29.1.1): 56 data bytes > ping: wrote 134.29.1.1 64 chars, ret=-1 > ping: wrote 134.29.1.1 64 chars, ret=-1 > ping: wrote 134.29.1.1 64 chars, ret=-1 > > --- 134.29.1.1 ping statistics --- > 3 packets transmitted, 0 packets received, 100% packet loss This is a different matter. This is typical of a down link. You send off packets to the name server, and for some reason they don't get sent. After 50 packets accumulate, the driver refuses to accept any more, and that's the message you get. > The output from mailq (when stuck is ): > Mail Queue (3 requests) > --Q-ID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------ > UAA12735 51 Tue Dec 3 20:09 pabraham > (host map: lookup (canes.gsw.peachnet.edu): deferred) > Elizabeth Abraham > <eabraham@canes.gsw.peachne > TAA12414 351751 Tue Dec 3 19:24 pabraham > (host map: lookup (epsilon.cs.mankato.msus.edu): deferred) > "David J. Haglin" > <haglin@epsilon.cs.mankato. > TAA12407 522300 Tue Dec 3 19:20 pabraham > (host map: lookup (epsilon.cs.mankato.msus.edu): deferred) > "David J. Haglin" > <haglin@epsilon.cs.mankato. Yes, that's reasonable. sendmail is attempting to look up the address of the recipient machines, and it can't get the name server machine. > I also note that this has instead of "lookup" "I/O error" after several > attempts. Interesting. I haven't seen that one, but it doesn't look important enough to follow up on yet. > I guess if I can increase the buffer space I will be able to solve the > problem. The output of df is: > Filesystem 512-blocks Used Avail Capacity Mounted on > /dev/sd0a 63550 25682 32784 44% / > /dev/sd0s2f 3855822 745712 2801646 21% /usr > /dev/sd0s2e 59454 12948 41750 24% /var > procfs 8 8 0 100% /proc The buffer space is in memory, not on disk. But you're not short of buffer space (the netstat says you're only using 24%), so there's no worry there. The real problem is that your ethernet link appears to be going down. There were problems with the vx driver in 2.1.5, and at a rough guess I'd say that's what's biting you. I don't know if 2.1.6 is any better, but you could try it, or if you're prepared to try the 2.2 snapshot, that might help. 2.2 will be coming out in a month or two, so if you can hold through until then, that might be a better alternative. Greg
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199612051621.RAA25361>