Machine says not enough space

[Brian_71]

/usr/lib/dld.sl: Call to mmap() failed - TEXT /cpapps/vis/u02/oracle/vis/visdb/8.1.7/javavm/admin/libjox8oracle_aurora_rdbms.sl
/usr/lib/dld.sl: Not enough space

Get this same error with a variaty of programs I try and run after the system has been up for a while. Any ideas

[Uday_S_Ankolekar]

Hello,

Check with bdf if any file system is full. I think u02 is filled up. But bdf command would show you the status of filesystem

-USA..

[PIYUSH D. PATEL]

Hi,

Check with bdf......your filesystem may have got full.
Make space and run the command again.

Piyush

[Martin Johnson]

You may also want to check memory too. vmstat or glance.


Marty

[Jeff Schussele]

Hi Brian,

I suspect that you've grown a process to the kernel's max allowed size - increase maxdsiz
See the following thread:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x26e9d211e18ad5118ff10090279cd0f9,00.html

HTH,
Jeff

[Helen French]

You can check the following to find out a possible solution:

1) swap space usage (swapinfo, gpm)
2) file system usage (bdf)
3) Permission of the specific file TEXT (chmod, chown permissions)
4) System patch level

[Helen French]

One more option:

Check the kernel parameters, especially 'maxdsiz' parameter.

[S.K. Chan]

Usually this is indicative of ..
a) filesystem full (use "bdf")
b) not enough swap space (use "swapinfo -tam")
c) kernel parameter (maxdsiz) set too low (use "glance" to check)

[PIYUSH D. PATEL]

Hi,

it could be something related to memory or swapspace also. Give the statistics of
swapinfo and check the memor utilisation using vmstat, glance and top.

Piyush

[Jeff Schussele]

Oh..if you're running the OS & Oracle 64 bit, then the kernel parameter would be maxdsiz_64bit.

Rgds,
Jeff

[MANOJ SRIVASTAVA]

Hi Brain


What does dmesg say , check /var/adm/syslog files too, sar -v o/p and also the Kernal Parameter SHMMAX too .



Manoj Srivastava

[Brian_71]

sar -v gives error
snoopy:/var/adm #sar -v
sar: Can't open /var/adm/sa/sa06

bdf shows all file systems are under 60 %

how do I find out the % shared mem being used.

No error were found in syslog.log

[S.K. Chan]

Like so .. ( example : 2 samples 5 seconds apart) ..

# sar -v 5 2

From the output..

Value-----KernelParameter
==========================
"proc-sz" (nproc)
"inod-sz" (ninode)
"file-sz" (nfile)

How's your swap doing ? You haven't mention that. Use "ipcs -ma" to look at your shared memory segments.

[Brian_71]

the ipcs -ma spits out a whole bunch of info.
How do I find out how much is used and how much is free.
We are running a 32 bit version of oracle 8i.
Hear some stuff on memory windows with HP.
Is this an HP specific issue?

[Bill Hassell]

The "out of space" message associated with dld.sl is always memory, not disk. It refers to the process'es ability to continue growth and fails due to:

1. maxdsiz has been exceeded

2. swapspace is full

For #1, run SAM and change maxdsiz to 950 megs. Then and reboot is complete, run the program again and it should be OK.

For #2, you've got big problems, namely: not enough RAM. Normally swapspace is casually occupied 5-20% for typical environments, but if swapspace is full, your RAM is WAY too small. For Oracle, you should think a minimum of 2Gb and ideally 4Gb for 32bit versions. For 64bit versions, the minimum is 4Gb RAM.

ipcs doesn't give you very much data and refers only to shared memory elements (memory, semaphonres, etc). Shared memory in 32bit apps is a part of a single address space which all 32bit processes must use (if they need shared memory). You probably need to read the proc_mgt amd mem_mgt white papers in /usr/share/doc, and also get a copy of shminfo from:

ftp://contrib:9unsupp8@hprc.external.hp.com/sysadmin/programs/shminfo/

As you will see, 32bit apps have significant limitations.