Disk file operations I/O

Bug 10361651  File open may hang/spin forever if dNFS is unresponsive

 This note gives a brief overview of bug 10361651. 

 The content was last updated on: 10-NOV-2011
 Click here for details of each of the sections below.

Affects:

Product (Component)	Oracle Server (Rdbms)
Range of versions believed to be affected	Versions BELOW 12.1
Versions confirmed as being affected	11.2.0.2 11.2.0.1
Platforms affected	Generic (all / most platforms affected)

Fixed:

This issue is fixed in

12.1 (Future Release) 11.2.0.3 (Server Patch Set)

Symptoms:	Related To:
Hang (Process Hang) Hang (Process Spins) Waits for "Disk file operations I/O" Stack is likely to include kgodm_identify	Direct NFS (dNFS)

Description

An Oracle process may hang indefinitely when trying to open a 
file on dNFS if dNFS is unresponsive for some reason when 
it should really timeout after a while and report an error 
that the file cannot be opened. The problem is that a timeout 
counter may wrap and so the open operation may never time out 
when it should.
 
Rediscovery Notes:
  The session shows as waiting for "Disk file operations I/O".
  The call stack shows kgodm_identify near the top
   and may include kgodm_getnfscdata above this on
   some platforms.
   eg: A typical substack might be:
        __nanosleep <- kgodm_identify <- odm_identify
         <- ksfdopn1 <- ksfdopn
  The session is using some small CPU as it repeatedly sleeps
   in a loop for very short intervals.
 
Workaround
  Address the underlying dNFS issue.
  Kill the stuck process.