[daip] TV Problem at UMRAO site

[Hugh D. Aller]

Dear AIPS support,
    I have (almost) completed an installation of 31DEC99 (I used a
July 1, 2000 image) on a set of sun workstations here at Univ. of Michigan.
I regestered them under EOS, the system on my desk.  Everything appears
to work but...  the TY on 4 of the 5 machines does not come up! 
    I did the build on our department server, an Ultra 1, and added 4 
other servers (two Ultra 1's and two Ultra 5's); EOS was one of the other
Ultra 1s.  Simply put: everything appears to work fine on the department
server (where the build was done). AIPSTV, MSGSRV, TEKSRV, TPMON etc all 
come up. However on any of the other machines the aips login messages are 
essentially the same until it sends the line:
XAS: ***  Using shared memory option for speed ***

on RA (working) dept server) one gets messages like:

XASERVERS: Start TV LOCK daemon TVSERV on ra
TVSERVER: Starting AIPS TV locking, Inet domain
XASERVERS: Start XAS on ra, DISPLAY :0.0
XAS: ***********************************
XAS: **  No TrueColor found
XAS: **  Resorting to PseudoColor
XAS: ***********************************
XAS: Using screen width height 1142 800, max grey level 196
XAS: ***  Using shared memory option for speed ***
XASERVERS: Start graphics server TEKSRV on ra, DISPLAY :0.0
AIPStek:  unable to open font "Huge", trying "fixed"....
XASERVERS: Start message server MSGSRV on ra, DISPLAY :0.0
AIPSmsg:  unable to open font "Huge", trying "fixed"....
     AIPS TV, TEKSRV & MSGSRV all appear
on EOS (one non-working example) one gets:
XASERVERS: Start TV LOCK daemon TVSERV on eos
TVSERVER: Starting AIPS TV locking, Inet domain
XASERVERS: Start XAS on eos, DISPLAY :0.0
XAS: ** TrueColor FOUND!!! 
XAS: Visual not = default, create 24-bit colormap
XAS: Using screen width height 1270 924, max grey level 255
XAS: ***  Using shared memory option for speed ***
Shared memory id failure: Invalid argument    <--- error message appears
XASERVERS: Start graphics server TEKSRV on eos, DISPLAY :0.0
AIPStek:  unable to open font "Huge", trying "fixed"....
XASERVERS: Start message server MSGSRV on eos, DISPLAY :0.0
AIPSmsg:  unable to open font "Huge", trying "fixed"....
     TEKSRV & MSGSRV appear but no TV
.............................
Our system manager claims that all machines have an identical kernels (i.e. the
IPCSHMEM option was invoked).  Services sssin, ssslock are present on all
machines (which I think is proven by TEKSRV and MSGSRV appearing).  By the
way TV=local gives the same behavior.
   I am aware of only three unique diferences between the ra (the dept server)
and the other machines:
 1. the original build was done on ra
 2. ra has 256MB of memory the other ultras only have 128MB
 3. the file system resides on ra (although every thing is nfs mounted anyway).
...........................
I have inspected all the scripts I can think of (starting with INSTEP1.LOG)
I reran FILAIP and POPSGN on EOS in case something wasn't initiated correctly
   during the initial build.
Everything appears to point at XAS, so I rebuilt it on EOS (from the Makefile
  in 31DEC99/Y/SERVERS/XAS) using gmake and placed it in $LOAD.  The result
  was the same. There may still be some environmental variable that is set 
  wrong; but I didn't notice anything obvious 
  
Incidently, we had this same problem in the 15OCT98 build that we did, but
  the graduate student involved in that build (Jose Salgado) was using an
  XTERM into the main server.  He also worked on this problem. but opted tp
  finish his thesis first!  We no longer have the option of that work-around,
  there are 6 of us using (or trying to use) AIPS and the Dept server won't
  stand up to a bunch of IMAGR tasks running while still acting as the central
  server.
.................................
Help!  Do you have any suggestions?  I am sure there is is something dumb
that we have (or have not) done; but we have run out of ideas.  I wonder
if I should just try a rebuild on another machine, but I suspect
that I would just make the same mistake.  If it would help, I would be
quite happy to phone you the aips login info, but perhaps there is something
very obvious.
Thanks in advance.
Sincerely,
hugh
Hugh D. Aller
Department of Astronomy
University of Michigan