[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20000207: solaris for intel, SHMMAX

>From: alan anderson <address@hidden>
>Organization: St. Cloud State
>Keywords: 200002072008.NAA25577 McIDAS-X Solarisx86


>Have hit a snag while installing solaris for intel on another
>Gateway machine.  Already have done this a few times, on identical
>machines, so know it can work.
>During the install, there were no statements about any part of the disk
>being bad, and I chose the install type as Initial which is supposed to 
>overwrite earlier files on the disk.


>After completing an install that showed no problems, I continued with
>the segment under the title 'preparing the workstation' and the 
>section dealing with shared memory.  


>Edited the /etc/system file with the forceload and set statements as directed.

Are you sure that there were not typos?

>The command /usr/sbin/sysdef | grep SHMMAX  still came back empty, 

It should look something like:

 268435456      max shared memory segment size (SHMMAX)

>so I looked at the output of 'sysdef'.  Found quite a few lines, but
>very little after the labeled section 'Tunable Parameters'.  Only about
>2 or 3 blank lines after that point, with the last one reading
>       Segmentation Fault

I would drop back and do the following:

o remove (or comment out) the lines you added to /etc/system
o reboot
o refun sysdef and see if you still get the Segmentation Fault

>Checking another twin machine, sysdef returns lots of stuff after the 
>Tunable Parameter section, including lines about shared memory and 
>the SHMMAX statement.

This is as it should be.

>So, what is the line  Segmentation Fault telling me ?

It is like sysdef is dying, but I can't be sure.  Is this machine accessible
over the net?

>This was the 2nd install of solaris on this machine, as we had a failed 
>make when trying to build mcidas.

What was the indication for the failed make?  What prompted you to reinstall
the operating system?

>I suspect that the  SHMMAX statement
>had not been checked after the initial edit of /etc/system, so we thought
>we had set the shared memory, but it had not worked, just as at present. 

On the previous installation of Solaris?  Shared memory is not needed to
build McIDAS-X.  It is only needed to run McIDAS-X sessions with more
than one image window.

>Thats all I can think of for background on this machine for now.

Would it be possible to get a login to this box (and get root privilege)?
Perhaps there is something wrong with the sysdef program; perhaps something



>From address@hidden  Tue Feb  8 07:40:32 2000

>I am sure I had the forceload and set commands typed correct when I 
>edited etc/system.

>I do not recall why the make failed, pretty sure we had the gcc and f2c 
>all in the right places.

>At any rate, the machine is met56D,,
>but I did not finish the network configuration stuff after the install.
>I will work on this today, but won't be done right away as I have a class.
>Will  try and have network config done by 1 pm our time.


>From address@hidden  Tue Feb  8 10:26:10 2000

Hi Tom

>I finished the network configuration items and met56D can 
>connect to the outside world 


>From address@hidden  Tue Feb  8 17:12:03 2000
>To: alan anderson <address@hidden>
>cc: address@hidden,
>        "Mike Schmidt" <address@hidden>
>Subject: 20000208: sysdef core dumps on St. Cloud State Solarisx86 box (cont.)


After thrashing around on your machine, I called in an expert, Mike Schmidt.
Here is what he had to say/did:

>This is a known problem on some installations.  I have applied patch
>107129-01 to Alan's system and sysdef seems to work.  The description
>contained in the patch is as follows;
>> Patch-ID# 107129-01
>> Keywords: sysdef SEGV modctl MODGETPATHLEN
>> Synopsis: SunOS 5.7_x86: /usr/sbin/i86/sysdef patch
>> Date: Jan/08/99
>> ...
>> Problem Description:
>> 4187740 sysdef command causes SEGV in x86 machine


just rebooted Alan's machine and have verified that sysdef works as you

/usr/sbin/sysdef | grep SHMMAX
 536870912      max shared memory segment size (SHMMAX)

and that shared memory is setup correctly.



>From address@hidden  Wed Feb  9 09:41:33 2000


>Thank you for providing the patch on our machine.  
>One more question.  Do you have any idea as to why 
>this problem occured, especially since we have installed 
>the same software (from the same CD) on several other
>machines, some identical in terms of hardware to this last.

It is likely that the machines are not 100% identical.  Like Mike said
in his earlier not, the sysdef problem was experienced on certain
hardware configurations and not others.

>Again, thanks for the great service.

You are welcom.


NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.