Re: what is "kfk: async disk IO"?
- From: Dion Cho <ukja.dion@xxxxxxxxx>
- To: martin.a.berger@xxxxxxxxx
- Date: Fri, 10 Dec 2010 15:18:49 +0900
As Freek has pointed it out, "kfk" is related to asynch I/O. Exactly speaking, kfk layer is activated by disk_asynch_io parameter being true. If the problem persists, strace on the process would give additional info about the asynchronous I/O system call. And you can get parameter info from V$EVENT_NAME as following. TPACK@ukja1120> exec print_table('select * from v$event_name where name like ''kfk'''); EVENT# : 377 EVENT_ID : 1568594048 NAME : kfk: async disk IO PARAMETER1 : count PARAMETER2 : intr PARAMETER3 : timeout WAIT_CLASS_ID : 4108307767 WAIT_CLASS# : 9 WAIT_CLASS : System I/O ================================ Dion Cho - Oracle Performance Storyteller http://dioncho.wordpress.com (english) http://ukja.tistory.com (korean) http://sites.google.com/site/otpack (tpack) ================================ On Tue, Dec 7, 2010 at 7:50 PM, Martin Berger <martin.a.berger@xxxxxxxxx>wrote: > Freek, > > thank you for your clarification. > Currently I'm not so much on the performance tuning side. It's more a kind > of troubleshooting, where I still does not know how to catch the situation. > Some minutes after my observation the instance hang as all sessions where > used - not even sys could logon. (yes, prelim worked, but that's not the > issue here). > No, my colleague did not create a systemdump or hanganalyze or similar.... > > Even it's only a 'test' system we had to restart the instance. Without any > big investigation. > Now I'm searching what this "kfk: async disk IO" is. Maybe someone has some > infos about P1, P2, P3. I will try to grab some informations out of ASH or > other resources, based on this. > > Martin > > Am 07.12.2010 um 10:42 schrieb D'Hooge Freek: > > > Martin, > > > > The "kfk: async disk IO" seems to be linked to the io_submit system call. > > You could check your number of async io processes. > > > > But don't look at the number of occurrences to determine how "heavy" an > event is, look at the time spend. There is no relation between number of > occurrences and time spend. > > > > > > Regards, > > > > > > Freek D'Hooge > > Uptime > > Oracle Database Administrator > > email: freek.dhooge@xxxxxxxxx > > tel +32(0)3 451 23 82 > > http://www.uptime.be > > disclaimer: www.uptime.be/disclaimer > > -- > > From: oracle-l-bounce@xxxxxxxxxxxxx [mailto: > oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Martin Berger > > Sent: dinsdag 7 december 2010 9:50 > > To: Oracle-L Freelists > > Subject: what is "kfk: async disk IO"? > > > > Hi, > > > > can anyone please hint me to some information about > > kfk: async disk IO ? > > > > A little query on v$sessions gave me > > USERNAME EVENT > COUNT(*) > > ------------------------------ > -------------------------------------------------- ---------- > > APPS ges message buffer allocation > 98 > > APPS CSS operation: action > 2 > > APPS Streams AQ: waiting for messages in the > queue 9 > > SYS SQL*Net message to client > 1 > > APPS SQL*Net message from client > 147 > > APPS kfk: async disk IO > 1646 > > > > as kfk: async disk IO is the bigest contributor, I'd like to investigate > a little bit. > > > > for the records: 18.104.22.168 - 2 node RAC - RedHat 5.4 Linux 64-bit. (yes, > it's EBS - but don't ask me the exact pach level, please - I assume > 11.5.10???) > > > > thnx > > Martin > > -- > http://www.freelists.org/webpage/oracle-l > > >