arndt@ZYX.SE (Arndt Jonasson) (06/21/89)
We have a number of workstations connected via ethernet and with each others' disks mounted via NFS. One of them is an HP9000/320 with HP-UX 6.2 running diskless, swapping on a HP9000/350. A third one is a DecStation 3100, running Ultrix 3.0. When I run the X11 window system on the 320's console, and access the DecStation through an xterm (running on the DecStation), the following problem sometimes occurs: When a lot of output appears at one time in the Ultrix xterm, the 'netisr' daemon running on the 320 starts to output lines looking like this on its console: June 21 ..time.. Network NetISR Error 2002 pid [ICS] ** Unknown log event These lines are sometimes repeated only some dozen times, sometimes they don't stop until after ten minutes. Nothing I do seem to determine when they stop. In particular, exiting the xterm window on Ultrix and umounting its disks from the 320 doesn't work. While these log messages are output, keyboard response on the 320 is very sluggish. The logging messages don't appear on any HP machine on the net except the one I am running the X11 server one. Not only xterm, but also Gnu Emacs 18.51 exhibits this problem. Does anyone have an idea what these messages mean, and what can be done about them (except not running Ultrix applications on a HP-UX X11 server :-( )? While they last, my workstation is next to unusable. There doesn't seem to be a comp.sys.ultrix, which is why I'm sending this to comp.unix.questions as well. Thanks for any help. -- Arndt Jonasson, ZYX Sweden AB, Styrmansgatan 6, 114 54 Stockholm, Sweden email address: arndt@zyx.SE or <backbone>!mcvax!enea!zyx!arndt -- Arndt Jonasson, ZYX Sweden AB, Styrmansgatan 6, 114 54 Stockholm, Sweden email address: arndt@zyx.SE or <backbone>!mcvax!enea!zyx!arndt
burdick@hpindda.HP.COM (Matt Burdick) (06/22/89)
> We have a number of workstations connected via ethernet and with each > others' disks mounted via NFS. One of them is an HP9000/320 with HP-UX > 6.2 running diskless, swapping on a HP9000/350. A third one is a > DecStation 3100, running Ultrix 3.0. > June 21 ..time.. Network NetISR Error 2002 pid [ICS] > ** Unknown log event > Does anyone have an idea what these messages mean, and what can be done > about them (except not running Ultrix applications on a HP-UX X11 server > :-( )? While they last, my workstation is next to unusable. The netisr process (/etc/netisr) on your 320 is not keeping up with the inbound LAN traffic. When this happens, the queue holding the inbound packets overflows and the error messages you've seen get logged to the console (usually screwing up the X display). 1) Check if netisr is running. If not, start it up (/etc/netisr 127). 2) Check if any other processes on your 320 are running at real-time priority. If so, it may be hogging the CPU so that netisr can't keep up. You could either remove real-time priority from the other process or processes or boost netisr's priority. Normally, netisr runs at real-time priority 127, which is the lowest available. It may be neccesary to boost it to 50 or more. 3) If all of this fails, consider masking the messages from appearing on your console. Use "netlogstart -cC" (I believe) to mask off the messages from the console. It may also be wise to either keep an eye on disk space or mask off logging to /usr/adm/neterrlog as well. -- Matt Burdick | Hewlett-Packard burdick%hpda@hplabs.hp.com | Technical Communications Lab
dwall@hpindda.HP.COM (Darren Wall) (06/23/89)
> We have a number of workstations connected via ethernet and with each > others' disks mounted via NFS. One of them is an HP9000/320 with HP-UX > 6.2 running diskless, swapping on a HP9000/350. A third one is a > DecStation 3100, running Ultrix 3.0. > June 21 ..time.. Network NetISR Error 2002 pid [ICS] > ** Unknown log event > Does anyone have an idea what these messages mean, and what can be done > about them (except not running Ultrix applications on a HP-UX X11 server > :-( )? While they last, my workstation is next to unusable. Matt Burdick gives some suggestions as to how to fix the problem ... however, you may want to a little more searching to find the root of the problem. We've seen the same problems here, and finally tracked it down to a broadcast storm occurring on our LAN. It turns out that another vendor's rwho product has a bug in it that causes it to forward the rwho broadcasts. This had the effect of causing a storm of messages to appear on the LAN every time an rwhod broadcasted it's information. Our solution was to not run rwhod, and everything has been working fine since then. Darren Wall Information Network Division of the Hewlett-Packard Company
diamant@hpfclp.SDE.HP.COM (John Diamant) (06/23/89)
> Does anyone have an idea what these messages mean, and what can be > done about them (except not running Ultrix applications on a HP-UX X11 > server :-( )? While they last, my workstation is next to unusable. I have no idea what the message means (besides the obvious literal interpretation), but I suspect you can stop the messages by changing your network logging mask on the console. Look at the -c option to netlogstart. I use -cD (log only Disaster level errors). I'm a lot happier with it now that it doesn't tell me every time I lose an RFA connection. Since I don't know the severity of that particular message, I can't be sure that it would disappear. John Diamant Software Engineering Systems Division Hewlett-Packard Co. ARPA Internet: diamant@hpfclp.sde.hp.com Fort Collins, CO UUCP: {hplabs,hpfcla}!hpfclp!diamant