commands are killed immediately.

[desbo]

I am in the process right now of changing over from a p630 to a 560Q. Everything has gone OK but when I sign on as one of the users and execute a command it is killed immediately.
We are running AIX 5.2 ML 8.
I have found something on the web about problems when ulimit is set to unlimited. It is set to unlimited but so was our old p630.

Can anyone help me with this quickly as it is nearly 20:00 and i am keen to get it finished and home.

Thanks

 

[p5wizard]

Can you check the error log?


HTH,

p5wizard

 

[desbo]

there are no errors displayed ion the error log

 

[p5wizard]

Any error messages when the processes get killed?

What commands do you run? Native AIX commands or an application?

Messages on the system console?


HTH,

p5wizard

 

[desbo]

It seems that AIX native commands work OK

No error messages

Example

990 BATH-adm1-> sqlplus
Killed

Current Directory is "/TEMP"
991 BATH-adm1->

Nothing on the console

 

[p5wizard]

OK. Some more Qs.

Can you run sqlplus as oracle user? If so, compare user environment with oracle user environment.
Is AIO available?
lsdev -C|grep aio
Is swap space big enough? (compare with old server)
lsps -a
Any errors in ORACLE alert log file?



HTH,

p5wizard

 

[desbo]

This box does not have oracle on it just oracle client. i can't run COBOL programs either so it is not just an oracle problem.

AIO - I get this displayed
1188 BATH-Root-> lsdev -C|grep aio
aio0 Defined Asynchronous I/O (Legacy)
posix_aio0 Defined Posix Asynchronous I/O

Swap space - HUGE!!!

No errors - it just doesn't even load the program

 

[p5wizard]

The aio shouldn't be a problem - only if it is ora db server, which it isn't.

Can you tnsping the database?

Is the DB server reachable/pingable from the new server?

Is tnsnames.ora available (/etc or somewhere in the oracle client network/admin dir)

You say swap space is huge? But is it also active?
I.e are there one ore more swapspaces? auto on?
show lsps -a please

Is the COBOL runtime installed? Perhaps a license issue?

Same oslevel as on the old server?
Same kernel mode? 32bit/64bit?

I'm running out of ideas here...


HTH,

p5wizard

 

[desbo]

It was the aio

It was available on our old box and just defined on the new one. I enabled aio and it all works fine now.

Thanks for your help

 

[p5wizard]

OK then,

hope you get home in time for dinner ;-)


HTH,

p5wizard

 

[desbo]

The dinner is in the bin already!!!

 

[p5wizard]

Did you configure aio with same attributes as on old server? You might introduce a perf issue if not...

number of aio servers
number of requests

I generally configure min 100, max 100 per CPU, 16000 requests. But that is on a DB server


HTH,

p5wizard

 

[RMGBELGIUM]

how about the maxuproc ? max number of processes per user ...
lsattr -E -l sys0 -a maxuproc

rgds,

R.

 

[desbo]

in reply to the previous two posts.

1. I never even knew about aio before yesterday.
On my previous box it is set at what I assume to be the defaults, min=1, max=10, requests=4096, server priority=39.
It is the same on my new box.
The machine is a very busy applications box running our ERP systems.
What is the impact of low values for aio and what does it actually do?

2. Thanks for reminding me about max user processes. It was not causing my problem yesterday but it did cause a problem this morning.

 

[khalidaaa]

How Many AIO Servers Do I Need?
Here are some suggested rules for determining what value to set maximum number of servers to:

The first rule suggests that you limit the maximum number of servers to a number equal to ten times the number of disks that are to be used concurrently, but not more than 80. The minimum number of servers should be set to half of this maximum number. 
Another rule is to set the maximum number of servers to 80 and leave the minimum number of servers set to the default of 1 and reboot. Monitor the number of additional servers started throughout the course of normal workload. After a 24-hour period of normal activity, set the maximum number of servers to the number of currently running AIOs + 10, and set the minimum number of servers to the number of currently running AIOs - 10. 
Note:
In some environments, you might see more than 80 AIOs kprocs running. If so, consider the third rule that follows.

A third suggestion is to take statistics using vmstat -s before any high I/O activity begins, and again at the end. Check the field iodone. From this you can determine how many physical I/Os are being handled in a given wall clock period. Then increase the maximum number of servers and see if you can get more activity or event completions (iodones) in the same time period.

http://publib.boulder.ibm.com/infoc...ix.kernelext/doc/kernextc/async_io_subsys.htm

Regards,
Khalid

 

[p5wizard]

aio is a method for offloading disk IO from the application to the operating system so that the application can proceed to the next transaction. I find it a bit strange that it should need to be available on non-dbserver machines.

If you have the same setup as for your previous server, you are more than likely OK with that.

Might well be that the oracle client software relies on aio being available, but doesn't actually use it.

run this:
ps -efk|grep aioserver
if all aioserver processes are consuming CPU time, then you probably don't have enough of them.
if there are aioserver processes still at 0:00, you should be fine.


HTH,

p5wizard

 

[desbo]

Dumb question but how do I monitor the number of additional servers started throughout the course of normal workload?
Thanks

 

[desbo]

On my applications machine there is just the one aio server process and it is not clocking any time. On my oracle server there are 40 processes and they are all clocking time. The oracle server is also set up with the default values as I stated before.

This machine has 4 cpus and has the maximum set to 10 per cpu and has 40 aio servers running so I would guess that the maximum is set to too small a value.

 

[p5wizard]

look at the values.

on app server, as I guessed, it just needs to be there, or you would have ended up with the max of 10 times number of CPUs.

On DB server, with default settings you started out with 1 and ended up with the max of 4x10, you can probably find out with ps -efk, when the additional aioservers got fired up, I'll bet it would be pretty soon after the DB got started.

Change the settings to 100/100/16000 - but you need to reboot the DB server for that to become effective.

again use ps -efk|grep aio to check if you have 'idlish' aioserver processes like a week after the DB server has restarted.

If it's a big DB server, you might even have to increase to 200/200/32000


HTH,

p5wizard

 

[desbo]

how much memory/processing power does increasing these values take? We have some processor capacity but practically no memory capacity free.
Thanks