AIX

Hai

I am new to AIX, currently in our shop we use AIX 4.3.3 , where we have hosted the websphere 3.5 applicaiton on it. The AIX server is getting hanged very frequently, when analyse we found the swap memory is increasing once it reaches the memory 100 the system is totaly in hung state, so we have to power of the system and to make the swap memory to 1% then only the system is recovered. So can anybody suggest something for this. Or do we have any fix patches for this .

Page Space Physical Volume Volume Group Size %Used Active Auto Type
paging00 hdisk0 rootvg 1024MB 68 yes yes lv
hd6 hdisk0 rootvg 512MB 100 yes yes lv


Vipin

It sounds to me that system might be running out of virtual memory and is performing a core dump which takes some time and space. You may also be running out of disk on the entire box.

If you have the space try and double the virtual memory size or add more disk space.

You may also want to perform the lsps -a or topas commands to watch the memory usage etc while it is running. Check for an application dump.

- sorry for double post, content deleted, see below -

It would be interesting to know on which fix level you are. Issue

instfix -i | grep AIX_ML

to find out. Also find out what you dump device is and if it is big enough to hold the dump. Use

sysdumpdev

with various parameters to get the size, location and other parameters of the dump and dump device. Read the man page of sysdumpdev to understand the concepts behind the system dump.

In any case it is strongly recommended to use evenly sized swap devices if you have several of them: AIX uses a round-robin scheme to fill pages in swap devices and if they are not evenly sized one will run out of space before the other one(s).

bakunin

bakunin,

Thanks for the reply, I have enclosed the output for the what are the fix pack we have and sydump . Also can you suggest me how to make the swap memory even. We have recenlty increased our swap memory to 1024 mb but once the 512mb reaches the stage of 85 our websphere applicatioon server could not do perform any task. I suspect this is some memory leak. Can you say some soultion how to solve it.

instfix -i |grep AIX_ML

Not all filesets for 4.3.1.0_AIX_ML were found.
Not all filesets for 4.3.2.0_AIX_ML were found.
All filesets for 4.3.1.0_AIX_ML were found.
Not all filesets for 4.3.2.0_AIX_ML were found.
Not all filesets for 4.3.3.0_AIX_ML were found.
Not all filesets for 4330-02_AIX_ML were found.
All filesets for 4320-02_AIX_ML were found.
Not all filesets for 4330-03_AIX_ML were found.
Not all filesets for 4330-04_AIX_ML were found.
Not all filesets for 4330-05_AIX_ML were found.
Not all filesets for 4330-01_AIX_ML were found.
Not all filesets for 4330-06_AIX_ML were found.
Not all filesets for 4330-07_AIX_ML were found.
Not all filesets for 4330-08_AIX_ML were found.
Not all filesets for 4330-09_AIX_ML were found.
Not all filesets for 4330-10_AIX_ML were found

sysdumpdev
primary /dev/hd6
secondary /dev/sysdumpnull
copy directory /var/adm/ras
forced copy flag TRUE
always allow dump FALSE
dump compression OFF

Thanks and Regards
P.Vipin

The OS has a setting that only allows a certian percentage of swap space to be used before it is going to send a kill signal to the process that is trying to go over the percentage. If you perform a vmstat -v you will see that the max client percentage is probally 80%. Perform this command: errpt pipe that to more. I bet you will see an error about swap space along with software program terminated. Anyhow this means that you do not have enough virtual memory allocated for what you are trying to do on the box. Double it again, make sure that if you are using more than one device you keep them at a 1 to 1 ratio such as the previous post it appears that you have two devices that are used for virtual memory one is at 512 and one is at 1024 if this is correct allocate 512 more to the 512 device and try it again.

Let me know what you find.