OpenSolaris

Printable Version Enter a New Search
Bug ID 6675447
Synopsis NFSv4 client hangs on shutdown if server is down beforehand
State 10-Fix Delivered (Fix available in build)
Category:Subcategory utility:smf
Keywords rtiq_reviewed
Responsible Engineer Pavel Filipensky
Reported Against solaris_10u4
Duplicate Of
Introduced In solaris_10
Commit to Fix snv_98
Fixed In snv_98
Release Fixed solaris_nevada(snv_98) , solaris_10u7(s10u7_01) (Bug ID:2161444)
Related Bugs 6544130 , 6733798 , 6778894 , 6806659 , 6808574 , 6810247 , 6888549 , 5070755
Submit Date 14-March-2008
Last Update Date 24-September-2009
Description 
On NFSv4, client hangs on shutdown process if NFS server has been shutdown
first.  It happens when renew or delegreturn operation is issued from async
threads after the NFS server goes down. Then async threads are never killed 
and umount process blocked on them.

Once we fall to the situation, we need system reset or stop-a for recovery.
Work Around 
Disable delegation by setting NFS_SERVER_DELEGATION=off on /etc/default/nfs of NFS server.
Use reboot(1M) instead of shutdown(1M).
Comments 
this does not appear to work properly and as intended, 'unmountall -l' still does
attempts to unmount autofs triggered NFS mounts, see my update from 
Entry 6 frank.batschulat [2008-12-01 17:53] in bug 6544130.
Similar hang issue was reported on NFSv3 on different site, and they confirmed 
it's solved by fix of this CR. Because the fix is not NFSv4-specific but done 
on svc.startd code, I changed the category/subcategory of this CR to utility/smf.