Bug 656003

Summary:	/var busy during shutdown because nfsd threads still lurking around
Product:	[Fedora] Fedora	Reporter:	Ronald Wahl <rwahl>
Component:	nfs-utils	Assignee:	Steve Dickson <steved>
Status:	CLOSED DUPLICATE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	low
Version:	14	CC:	jlayton, steved
Target Milestone:	---
Target Release:	---
Hardware:	Unspecified
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2010-11-27 18:27:27 EST	Type:	---
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---

Description Ronald Wahl 2010-11-22 15:29:59 EST

Description of problem:
When rebooting or halting the system /var cannot be unmounted because it is busy. The reason is that the nfsd kernel threads are not terminated synchronously but with a delay. After the last of these threads is gone the message "nfsd: last server has exited, flushing export cache" appears in dmesg and the /var can be unmounted. But it takes some minutes until the last server exits which is too long for the shutdown process.

I tried to manually reproduce it and found that if the network is shut down before the last nfsd thread is gone it will hang. So it looks like a race condition. Probably the stop operation of the nfs start script should enforce that all nfsd threads are gone before returning. E.g. this can be done by calling:

rpc.nfsd 0

which immediately stops all nfsd threads.

Version-Release number of selected component (if applicable):
nfs-utils-1.2.3-1.fc14.x86_64

How reproducible:
Always

Steps to Reproduce:
1. start rpcbind and nfs service
2. stop network
3. shutdown nfs service
4. Verify that all nfsd threads are gone (ps ax|grep nfsd)
  
Actual results:
there are still nfsd threads

Expected results:
no nfsd threads

Additional info:

Comment 1 Ronald Wahl 2010-11-27 18:27:27 EST

Ah, it looks like this is a dupe of Bug 655726. So it looks like the nfs service is not stopped at all because of some generic code making assumptions that are not always true.

*** This bug has been marked as a duplicate of bug 655726 ***