KHF@JHUIGF.BITNET.UUCP (10/05/87)
Hi gang:
We're having a DECnet problem that has finally driven me around the bend.
We've got a pair of uVAX II's running uVMS 4.5/DECnet 4.5 (full routing).
One has 5 MB, the other 9 MB. On each machine, we are receiving what I
consider to be an unusual number of SYSTEM BUFFER UNAVAILABLE counts from
NCP SHOW LINE QNA-0 COUNTER --- 50 or more events PER HOUR!!! The NCP manual
explains the message as:
"This counter indicates the total number of times no system buffer was
available for an incoming frame. This can be any buffer between the hardware
and the user buffers (those supplied on receive requests)."
Pretty useful message, eh? Swapping DEQNA boards, changing the number
of receive buffers (SET LINE QNA-0 RECEIVE BUFFERS) from 6 to 12 each had no
effect. MONITOR POOL indicates that we are not running out of LRP's (or SRP's
and IRP's for that matter).
What's going on here? Any ideas? I should also add that these
machines are on an Ethernet cable with a large number of systems running TCP/IP
and XNS. Could that enter into it? Any help at all would be GREATLY
appreciated.
-------------------------------------------------------------------------------
Ken Fasman
U.S. Mail: Bard Laboratories of Neurophysiology
Johns Hopkins School of Medicine
725 N. Wolfe St.
Baltimore MD 21205
Telephone: (301) 955-8330
BITNET: KHF@JHUIGF
"My opinions, especially the wrong ones, are strictly my own."
-------------------------------------------------------------------------------klb@philabs.Philips.Com (Ken Bourque) (10/07/87)
In article <8710052221.AA23415@ucbvax.Berkeley.EDU> KHF@JHUIGF.BITNET writes: >Hi gang: > > We're having a DECnet problem that has finally driven me around the bend. >We've got a pair of uVAX II's running uVMS 4.5/DECnet 4.5 (full routing). >One has 5 MB, the other 9 MB. On each machine, we are receiving what I >consider to be an unusual number of SYSTEM BUFFER UNAVAILABLE counts from >NCP SHOW LINE QNA-0 COUNTER --- 50 or more events PER HOUR!!! It looks like you have a loading problem on the network. It could be that one or more nodes (not necessarily DECnet nodes) has a problem and is spraying out garbage packets. Or, maybe you have a genuinely heavy load (less likely). If you have an Ethernet monitor check the traffic and see if any of the nodes are transmitting heavily (esp. to broadcast or multicast addresses) - then disconnect them and see if your buffer problem goes away. Ken Bourque klb@philabs.philips.com ...!uunet!philabs!klb