Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 7 Nov 1999 22:26:06 -1000 (HST)
From:      Bill Marquette <billm@danger.ms>
To:        freebsd-current@FreeBSD.ORG
Subject:   No buffer space available errors
Message-ID:  <Pine.BSF.4.10.9911072157590.10718-100000@nuke.danger.ms>

next in thread | raw e-mail | index | archive | help
For the last week and a half or so I've been trying to track this down
assuming it was a configuration error on my part or a problem with my
ISP's DHCP configuration.  After switching from a DEC 20141 chipset card
to a 3com 3c905, I found I was still having problems although the error
message had changed.  I'm now back to the DEC card cause I found more
results in the mailing lists on the errors I was seeing.

At first it appeared to be dhclient having issues, but I've finally found
occurances of other programs showing symptoms prior to dhclient log
entries.  Also, until today I couldn't replicate the problem while sitting
at the console, it would only show itself after I'd been away for a few
minutes (I swear it has a mind of it's own) usually, when I was going to
be away for more than 10-15 minutes.  Today I managed to force the machine
to have problems by doing multiple simultaneous downloads and uploads, I
was running about 100K/sec through the NIC.

This machine is running as a simple workstation with nothing special on it
but there were alot of messages that mentioned increasing maxusers and/or
NMBCLUSTERS to solve this.  I've now got maxusers at 128 and NMBCLUSTERS
at 16384 (in fairly small increments to see where it would start working).
NBUF is also currently at 4096 again, raised over a period of two weeks at
a recommendation from a friend.  I'm still getting the "No buffer space
available" messages and the only fix is to kill dhclient, ifconfig down
the interface and restart dhclient.

System and kernel are of the same sources, cvsupped multiple times over
last two weeks.  Most recent from October 31st.  I will be performing a
cvsup again tomorrow and see if that makes a difference.

Thanks for any help!!!  The following is either excerpts from logs or
various utilities that have been used or asked for in other reports on
similar problems.

--Bill

Hardware:

AMD K6-2 400
192MB RAM
Voodoo3 2000 PCI video
2x Maxtor 8gb IDE
12x Mitsumi ATAPI CDROM
2x Memorex ATAPI CD/RW
AHA 1542 SCSI controller
4gb IBM Fast SCSI-2
32x Toshiba SCSI CDROM
Soundblaster AWE32 (using drivers from www.4front-tech.com)
3Com 3c905 10/100 NIC (currently unused while tracking down problem)
Generic DEC 2104x chipset 10Mbit NIC
Externel parallel port ZIP 100


From dmesg:
xl0: <3Com 3c905B-TX Fast Etherlink XL> irq 10 at device 8.0 on pci0
xl0: Ethernet address: 00:50:04:66:ef:a8
miibus0: <MII bus> on xl0
ukphy0: <Generic IEEE 802.3u media interface> on miibus0
ukphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
de0: <Digital 21041 Ethernet> irq 12 at device 9.0 on pci0
de0: 21041 [10Mb/s] pass 1.1
de0: address 00:80:ad:a9:30:f9


From syslog:
Nov  7 14:09:01 nuke named[153]: ns_forw: sendto([216.225.0.4].53): No
buffer space available
Nov  7 14:09:30 nuke named[153]: ns_forw: sendto([206.85.240.238].53): No
buffer space available
Nov  7 14:09:59 nuke named[153]: ns_forw: sendto([128.9.0.107].53): No
buffer space available
Nov  7 14:10:00 nuke named[153]: ns_forw: sendto([209.117.223.51].53): No
buffer space available
Nov  7 14:11:00 nuke named[153]: ns_forw: sendto([206.239.111.254].53): No
buffer space available
Nov  7 14:11:39 nuke dhclient: DHCPREQUEST on de0 to 204.210.96.1 port 67
Nov  7 14:11:39 nuke dhclient: send_packet: No buffer space available

From netstat -m  (Note: I had been downloading some rather large files
and wasn't at console for a couple hours; this is highly exagerated since 
I've seen this error with as few as 300/900 mbufs in use):

45/34496 mbufs in use:
        38 mbufs allocated to data
        7 mbufs allocated to packet headers
36/16158/16384 mbuf clusters in use (current/peak/max)
36628 Kbytes allocated to network (0% in use)
0 requests for memory denied
0 requests for memory delayed
0 calls to protocol drain routines                            

sysctl vm.zone:

vm.zone:
ITEM            SIZE     LIMIT    USED    FREE  REQUESTS
socket:          192,    16384,    145,     86,    46856



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.10.9911072157590.10718-100000>