Connectivity abruptly lost

[mehul_3]

While application is running, connectivity to server from all client is was abruptly lost and all user was not able to access application due to oracle listner fail and I notified the following error in listener.log
TNS-12500: TNS:listener failed to start a dedicated server process
TNS-12540: TNS:internal limit restriction exceeded
TNS-12560: TNS:protocol adapter error
TNS-00510: Internal limit restriction exceeded
HPUX Error: 12: Not enough space
I restarted HP-Ux11i box resolving the issue but it may comes again in near futur.
Pl. suggest everlasting solution.

Note:3GBRAM,TotalSGA=1400MB and pga_aggregate_target=550MB

Waiting for immediate response as application is very critical(Billing).

Regards,
Mehul

[Devender Khatana · Accepted solution]

Hi,

The messaage here indicates that either swap space is fully utilizaed or some kernel parameter has been totally used. I would suggest to monitor both of these especially when next time the problem appears and when load on the system approaches maximum.

#swapinfo -atm ( To monitor swap)

#sar -v 1 11 ( To know parameter utilizations)

HTH,
Devender

[Steven E. Protter]

Shalom Mehul,

Now would be a good time to monitor performance.

http://www.hpux.ws/system.perf.sh

It's production code, runs background and collects and summarized data in many performance areas.

SEP

[Bill Hassell]

The critical error is 12 which is not enough RAM. The quickest fix is to add more swap space. Find or create an empty lvol and use swapon to add it. swapinfo -tam will probably report fairly high swap space usage if this is the problem. Otherwise, the problem likely due to shared memory fragmentation which will take a while to sort out (without accidently killing some programs). A reboot will fix this.

The 1400meg SGA is fairly large and any other shared memory usage (such as fbackup or other programs) which ran and then aborted, or you restarted the database without a reboot, these could cause the lack of available RAM. 32bit programs that use large amounts of shared memory are particularly vulnerable to memory map fragmentation (not a problem for 64bit programs). As long as the database is shitdown and restarted with no other special programs running, the map should remain intact.

[Jean-Luc Oudart]

Hi,

These Oracle parameters seem quite high regarding the overall available memory !
Did you (or your DBA) use the cache advisory to determine these values ?
If yes, there is probably only one solution for you as you mentioned the application is critical and therefore I think you have a business case to increase the amount of memory on this server.

If not, you may run it and compare values for SGA and PGA with the cache advisor ones.
(you may still want extra RAM for this server).

Regards
Jean-Luc