farhad@CS.Stanford.EDU (Farhad Shakeri) (07/20/90)
I do not know if this has been talked about but I missed it.
When I installed our 3100s I just NFS mounted the /usr form
out 5400 to save time and space. All of the binaries on
the 5400 were fine and we never had any problem with them. but
the moment I and other users tried to use these binaries:
X11R4 binaries, gnuemacs, tcsh, MH... they all acted
strange, from core dump, seg. faults to hangings really
annoying. Then I just recompiled everything on my 3100 and
all went back to normal. We are running the same OS 3.1c
on our 5400s and 3100. I complained to DEC and they are
telling me that 5400 binaries and 3100s are compatible;
a 5400 was made to act as a server for 3100s.
One more thing is that after I played around with the 5400
binaries on 3100, I can rate something like, they are
compatible %80 and %20 not compatible.
Also has anyone tried to make a 3100 kernel in /sys of
a 5400?
Or has anyone ever seen a problem like this? What am I
doing wrong?
Thanks for your help.
+----------------------------------------------------+
/ Farhad Shakeri E-Mail: /
/ Stanford University farhad@Tehran.Stanford.EDU /
/ Computer Science Dept. /
+----------------------------------------------------+grr@cbmvax.commodore.com (George Robbins) (07/21/90)
In article <1990Jul20.162904.3793@Neon.Stanford.EDU> farhad@CS.Stanford.EDU (Farhad Shakeri) writes: > I do not know if this has been talked about but I missed it. > > When I installed our 3100s I just NFS mounted the /usr form > out 5400 to save time and space. All of the binaries on > the 5400 were fine and we never had any problem with them. but > the moment I and other users tried to use these binaries: > X11R4 binaries, gnuemacs, tcsh, MH... they all acted > strange... I'm doing essentially the same thing with a 3100 and a 5800 with 3.1c and haven't noticed any problems of this sort. > Also has anyone tried to make a 3100 kernel in /sys of > a 5400? Yep, seems to work. Make sure the config file has all the appropriate entries for a "UWS"* system and device entries. Releases prior to 3.1[a-d] didn't use explict config file entries for DS3100 devices. I think there are some suitable sample config files in the 3.1c /sys/conf directory to work from. > Or has anyone ever seen a problem like this? What am I > doing wrong? I dunno, did you install 3.1c from tape on both systems or did you upgrade the 3100's from a different release? Are the 3100's as workstations with UWS 2.2 libraries loaded on the server? Did you install the mandatory patch tape? so on... -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
pavlov@canisius.UUCP (Greg Pavlov) (07/22/90)
In article <1990Jul20.162904.3793@Neon.Stanford.EDU>, farhad@CS.Stanford.EDU (Farhad Shakeri) writes: > > [ indicating problems in moving binaries from DEC 5400 to DEC 3100's ] We haven't tried to move system binaries. But we have a DEC 5400, DEC 5810, and DEC 3100's and have been able to move binaries betweenthem at will - including the INGRES DBMS package, BLAST, and a commercial text editor. Also, a large number of our own applications programs. greg pavlov, fstrf, amherst, ny
scs@iti.org (Steve Simmons) (07/23/90)
In article <1990Jul20.162904.3793@Neon.Stanford.EDU> farhad@CS.Stanford.EDU (Farhad Shakeri) writes: > When I installed our 3100s I just NFS mounted the /usr form > out 5400 to save time and space. All of the binaries on > the 5400 were fine and we never had any problem with them. but > the moment I and other users tried to use these binaries: > X11R4 binaries, gnuemacs, tcsh, MH... they all acted > strange... grr@cbmvax.commodore.com (George Robbins) writes: >I'm doing essentially the same thing with a 3100 and a 5800 with >3.1c and haven't noticed any problems of this sort. Hmm . . . DEC went out of their way to tell us *not* to run Ultrix 3.1c binaries on 3100s -- they didn't know of any problems, just said it hadn't been tested. Our experience between 3100s and 5810s has been 100% binary compatability. If we could compile and run it on one, the binary would work on the other. It's been a real plus to use a 3100 as a test bed for new software development. Lets us shake the bugs out in live use before putting it on the 5810 to screw up all the users. :-) Another local site with a 5400 and 3100s reports no problems with interoperability.
farhad@CS.Stanford.EDU (Farhad Shakeri) (07/24/90)
In article scs@iti.org (Steve Simmons) writes: |> [my stuff deleted]... |> grr@cbmvax.commodore.com (George Robbins) writes: |> >I'm doing essentially the same thing with a 3100 and a 5800 with |> >3.1c and haven't noticed any problems of this sort. |> |> Hmm . . . DEC went out of their way to tell us *not* to run Ultrix 3.1c |> binaries on 3100s -- they didn't know of any problems, just said it |> hadn't been tested. |> |> Our experience between 3100s and 5810s has been 100% binary compatability. |> If we could compile and run it on one, the binary would work on the other. |> It's been a real plus to use a 3100 as a test bed for new software |> development. Lets us shake the bugs out in live use before putting it |> on the 5810 to screw up all the users. :-) |> |> Another local site with a 5400 and 3100s reports no problems with |> interoperability. Well I do not remember if I said this before. Most of the binaries work fine except some of the X11R4 binaries and most of MH binaries. tcsh core dumped when it was called withing emacs and emacs (epoch?) acted abnormal. I have been told that there may be a version incompatibility, but I am runing the latest greates Ultrix 3.1c on all of my DEC (risk) machines. DEC is comming tomorrow and I am going to show them the problem. +----------------------------------------------------+ / Farhad Shakeri E-Mail: / / Stanford University farhad@Tehran.Stanford.EDU / / Computer Science Dept. / +----------------------------------------------------+
jg@crl.dec.com (Jim Gettys) (07/24/90)
As far as I know, (and we've run all sorts of Ultrix 3.1 and 4.0 variants on 3100's, 2100's, 5800's and 5000's here at CRL), you should have no trouble with any application except those which look at /dev/kmem and /dev, which have differences (as usual) between releases. This is the typical caveat for Unix systems in general. As usual, as well, no guarantees on downward compatibility between releases. We share a single /usr/local on all of our machines here with no problems, with both 3.1 and 4.0 represented. Someday (read OSF/1) /dev/kmem should be less of a problem when we have a tables system call... And the device numbers should not have to change again, either. - Jim
grr@cbmvax.commodore.com (George Robbins) (07/24/90)
In article <scs.648748265@hela.iti.org> scs@iti.org (Steve Simmons) writes: > In article <1990Jul20.162904.3793@Neon.Stanford.EDU> farhad@CS.Stanford.EDU (Farhad Shakeri) writes: > > When I installed our 3100s I just NFS mounted the /usr form > > out 5400 to save time and space... > > grr@cbmvax.commodore.com (George Robbins) writes: > >I'm doing essentially the same thing with a 3100 and a 5800 with > >3.1c and haven't noticed any problems of this sort. > > Hmm . . . DEC went out of their way to tell us *not* to run Ultrix 3.1c > binaries on 3100s -- they didn't know of any problems, just said it I should point out that DEC has been quite consistant in saying this, however when it come down to wasting ~150 M-byte of disk to serve an older version, or forking out $nK to special order 3.1D media which DEC hasn't otherwise seen fit to send me, then I'm reasonably happy to notice that it seems to work quite nicely. -- George Robbins - now working for, uucp: {uunet|pyramid|rutgers}!cbmvax!grr but no way officially representing: domain: grr@cbmvax.commodore.com Commodore, Engineering Department phone: 215-431-9349 (only by moonlite)
jg@crl.dec.com (Jim Gettys) (07/24/90)
Strange... We have X11R4 installed in our /usr/local/ area, and have no problems. Then again, I think we may have 3.1D installed on our 5800. We had several 5000's running 3.1D for a while and wanted to reduce the number of variants we were running. Certainly haven't seen tcsh or emacs bugs... - Jim