Need your helps,pls!

[FireDragon.Xu]

Recently the following problem has occurred on my computer: The client could not remotely access the system via telnet or FTP,though the system responded to ping. The system displayed the following error message while trying to loging on console:
/usr/lib/dld.sl: Can't allocate memory Strings
/usr/lib/dld.sl: Not enough space
/usr/lib/dld.sl: Call to mmap() failed - DATA /usr/lib/libxti.2
/usr/lib/dld.sl: Not enough space
/etc/profile[105]: 4436 Abort
None of the accounts could login. However it was possible to use rs command to reboot the computer in the GSP> prompt.. System returned to normal after rebooting. I would heartly appreciate any suggestions.

[Sridhar Bhaskarla]

Hi,

That could be mostly due to running out of swap space. Look at your swapinfo -t and see how much % was is being used.

You may want to add extra swap. Find out a volume group that is less used and add swap from it.

#lvcreate -n lvswap -L size_in_MB -r N /dev/vgxx

Edit /etc/fstab and add this entry

/dev/vgxx/lvswap ... swap pri=0 0 0

Run the command

#swapon -a

Run swapinfo -t and verify the extra device has been added.

-Sri

[Rajeev Shukla]

It looks like the system ran out of memory and could not assign memory to any process that does malloc() and as seen even mmap() failed. The best would have been to get your core dump diagnosed by HP response Team if you took so.
Other thing did you do any recent changes to kernel parameters.

Cheers
Rajeev

[Sridhar Bhaskarla]

Hi,

Just to give you little more information. When a process is spawned, the kernel will try to reserve space in the swap area. So if you are running out of swap area, then the new processes cannot be started and you get the error "Not enough Space".
There is a configurable parameter called "swapmem_on" and if it is set to 1, then the kernel will allot 75% of physical memory for reservation purposes so that you do not have to add unnecessary device swap.

If this kernel parameter is not 1, then you may want to enable it. Or if you can afford extra disk space, you can configure swap from it.

Keep an eye on the pageouts (po column under vmstat output). If they are too high, then your system may be under memory pressure.

-Sri

[Ravi_8]

Hi,

Your swap is filled, you need to increase the swap. also look at other file systems, Even if /tmp is filled, problems like this will occur

[Stanimir]

Hi!
As mentioned above, you need to assign
SWAPMEM_ON=1 - this will allow memory
to exceed swap space. Usually all processes
dont need all swap, reserved for them.
Anyway there could be other reason for
crash - please report row:105 of your /etc/profile.
Regards