Solaris 2.6 sysdef (vs. 2.5[.1])

Solaris 2.6 sysdef (vs. 2.5[.1])

Post by Andrew Gide » Tue, 15 Sep 1998 04:00:00



Under Solaris 2.6, the "Process Resource Limit Tunables" listed
in the result of sysdef seem to have changed format.  These are
either very large, or negative, hex values now.  In Solaris 2.5.1,
the values were either "normal" numbers or "Infinity".

How are lines like:

*
* Process Resource Limit Tunables (Current:Maximum)
*
ffffffff:fffffffd       cpu time
ffffffff:fffffffd       file size
ffffffff:fffffffd       heap size
ffffffff:fffffffd       stack size
       0:7ffff000       core file size
ffffffff:fffffffd       file descriptors
       0:  800000       mapped memory

to be interpreted?

Thanks...

        Andrew

 
 
 

Solaris 2.6 sysdef (vs. 2.5[.1])

Post by Casper H.S. Dik - Network Security Engine » Wed, 16 Sep 1998 04:00:00


[[ PLEASE DON'T SEND ME EMAIL COPIES OF POSTINGS ]]


>How are lines like:
>*
>* Process Resource Limit Tunables (Current:Maximum)
>*
>ffffffff:fffffffd       cpu time
>ffffffff:fffffffd       file size
>ffffffff:fffffffd       heap size
>ffffffff:fffffffd       stack size
>       0:7ffff000       core file size
>ffffffff:fffffffd       file descriptors
>       0:  800000       mapped memory
>to be interpreted?

Not readily.

The rlimit values prior to 2.6 were unsigned 32 bit integrals.

The introduction of largefiles in 2.6 made it necessary to expand
the values to 64 bits.

Unfortnately, sysdef reads the kernel directly and so no type checking
is done. (For 32 bit programs, the limit values still compile as 32 bit values)

So sysdef now reads completely bogus information.

The first line is the soft CPU limit, the second the hard CPU limit,
followed by soft/hard filessize, soft/hard heap limit and last the
soft stack limit.  The rest of the limits is not listed.

ffffffff:fffffffd is = (long long) -3 == RLIM_INFINITY

Casper

--
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

 
 
 

Solaris 2.6 sysdef (vs. 2.5[.1])

Post by Andrew Gide » Wed, 16 Sep 1998 04:00:00



Quote:

>Unfortnately, sysdef reads the kernel directly and so no type checking
>is done. (For 32 bit programs, the limit values still compile as 32 bit values)

>So sysdef now reads completely bogus information.

So what sysdef displays as two 32 bit values is in fact
one 64 bit value.  I understand.

Quote:

>The first line is the soft CPU limit, the second the hard CPU limit,
>followed by soft/hard filessize, soft/hard heap limit and last the
>soft stack limit.  The rest of the limits is not listed.

So is there some other way to get the information not listed
by sysdef?  Is it the same as that returned by ulimit?

Thanks...

        Andrew

 
 
 

Solaris 2.6 sysdef (vs. 2.5[.1])

Post by Casper H.S. Dik - Network Security Engine » Thu, 17 Sep 1998 04:00:00


[[ PLEASE DON'T SEND ME EMAIL COPIES OF POSTINGS ]]


>So is there some other way to get the information not listed
>by sysdef?  Is it the same as that returned by ulimit?

adb, I'm afraid.

In 2.7, there's "plimit 1" which should print what you want:

1:      /etc/init -
   resource              current         maximum
  time(seconds)         unlimited       unlimited
  file(blocks)          unlimited       unlimited
  data(kbytes)          unlimited       unlimited
  stack(kbytes)         8192            unlimited
  coredump(blocks)      unlimited       unlimited
  nofiles(descriptors)  64              1024
  vmemory(kbytes)       unlimited       unlimited

--
Expressed in this posting are my opinions.  They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth.

 
 
 

1. Solaris 8 ifconfig (lo0) problem VS Solaris 2.5,2.6,7

Under all versions of Solaris, except 8, when you ifconfig your loopback
address (lo0) with an IP address of some other host that host won't be able to
ping or make any network connections to the server.  That's obvious because the
ip address is now local to the machine and everything just breaks for that
host.

HOST = 10.20.20.200/24
Some router in between to connect each other.
Server = 10.2.0.2/24

Then you "ifconfig lo0:1 10.20.20.200 netmask 255.255.255.255 up" and the host
cannot talk to the server anymore.

In the Solaris 8 environment, even patched up with xref file 2/28/01, when you
do an "ifconfig lo0 addif 10.20.20.200 netmask 255.255.255.255 up" or "ifconfig
lo0:1 plumb ; ifconfig lo0:1 10.20.20.200 netmask 255.255.255.255 up" the host
is still able to talk with the server.  If you wait like 2 hours, sometimes the
host will not be able to ping/talk to the server until you unplumb lo0:1 then
traffic will start right away.

Has anyone else seen this?  Is this an "added feature" in solaris 8?  Any
suggestions to make it work like previous versions of Solaris.  Well for that
matter make it would like is should.   :)

-- Shawn

2. What happened to....

3. SunOS 5.6 vs Solaris 2.6 vs Solaris 7 vs Solaris 8

4. Token Ring/Netmask Problem

5. netstat on solaris 2.6 vs 2.5

6. Linux root user password lost

7. Solaris 2.5 vs 2.6 performance

8. Some advice needed

9. What happened to Solaris 2.6 sysdef

10. Sparc10 solaris 2.5 to Sparc20 solaris 2.6 migration question

11. Can software compiled for Solaris 2.5 be used in Solaris 2.6?

12. Solaris 2.5 ---> Solaris 2.6 -- won't let me upgrade

13. Benchmarks x86 Solaris vs Linux and Solaris 2.6 x86 vs sparc