25582 error on high volume of short live trans
Hi IIUG,
Trying to find why my DB blows up to high CPU. it has 62 physical cores (x4
smt) on AIX machine P8 running 11.70 IDS.
We are getting -25582 error after putting significant load of transactions.
The volume of transactions is big but short live.
Here's the config below:
NETTYPE ipcshm,1,100,CPU
NETTYPE soctcp,35,500,NET
LISTEN_TIMEOUT 10
MAX_INCOMPLETE_CONNECTIONS 1024
FASTPOLL 1
NUMFDSERVERS 23
NS_CACHE host=900,service=900,user=60,group=900ASF_SOCTCP_BACKLOG 1024
BUFFERPOOL
default,buffers=23068672,lrus=511,lru_min_dirty=0.28,lru_max_dirty=1.12
BUFFERPOOL
size=4K,buffers=69206016,lrus=511,lru_min_dirty=0.28,lru_max_dirty=1.12
We've tried NUMFDSERVERS=1 as well, but we are still seeing nsf locks. Any
thoughts?
Thank you in advance,
Hi Ronald,
I'm looking at this old post and probably the problem has already been solved.
I assume that when the 25582 error occurred there were also several nsf locks
records in the onstat -g lmx output.
I would say you should increase the NUMFDSERVERS parameter to 35 and try.
It would be interesting to know what the definitive solution was.
Best regards.
Roger