All of lore.kernel.org
 help / color / mirror / Atom feed
* lockd grace period
@ 2005-01-31  7:41 mehta kiran
  2005-01-31 12:33 ` Steve Dickson
  0 siblings, 1 reply; 13+ messages in thread
From: mehta kiran @ 2005-01-31  7:41 UTC (permalink / raw)
  To: nfs

Hi , 
        lockd is used by both nfs and autofs
if there are 2 machines runnnig 2.6 kernel
(Red Hat) .autofs and nfs both are running
on system1.When nfs is failed over to system2
, all daemons start on system2 , but
lockd on system1 is still alive as it is required
by autofs "client"(this machine has automounted
filesystem from some other machine).After say 5
minutes , if nfs
is failed over to system1 , lockd recovery
gives some problem.statd notifies client , 
but when client tries to reclaim lock , 
LOCK routine returns "unknown error"(30000).
Is is because  , lockd was runnnig since long
time on system1 and hence its grace period was over.

But I dont face same problem of sles9 sp1.
What is the expected behaviour.?
Is it that whenever nfsd starts  , lockd should enter
grace period ?
or
Is is that lockd should enter grace period only when
it restarts without caring for nfs.?

--thanks,
   kiran



	
		
__________________________________ 
Do you Yahoo!? 
Yahoo! Mail - You care about security. So do we. 
http://promotions.yahoo.com/new_mail


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

^ permalink raw reply	[flat|nested] 13+ messages in thread
* RE: lockd grace period
@ 2005-01-31 12:50 Aswathanarayana, Chetan (STSD)
  2005-01-31 14:36 ` mehta kiran
  2005-01-31 14:58 ` Steve Dickson
  0 siblings, 2 replies; 13+ messages in thread
From: Aswathanarayana, Chetan (STSD) @ 2005-01-31 12:50 UTC (permalink / raw)
  To: Steve Dickson, mehta kiran; +Cc: nfs

Steved wrote :

> Yes and Yes... When a NFS server is restarted, there is a grace period
in which the server (i.e. lockd) will accept reclaimed locks.
> And it appears that reclaimed locks will also be accepted after the
grace period expires=20

I think reclaim locks should not be accepted after the grace period
expires, because it can cause data corruption.

Say, a reclaim request arrives after grace period and get's lock. There
will be time frame between end of grace period and arrival of
Reclaim request, in which case any thing can happen(ex: a new request
could be granted, it may make some changes to a file and unlock by the
time Reclaim arrives).

In which case reclaim thinks that it was the same original file it had
last locked.

So reclaim after grace period is not a good idea.

Mehta's question :

>What is the expected behaviour.?
>Is it that whenever nfsd starts  , lockd should enter grace period ?
>or
>Is is that lockd should enter grace period only when it restarts=20
>without caring for nfs.?

The lockd should enter grace period whenever nfsd starts/restarts.

~Chetan


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

^ permalink raw reply	[flat|nested] 13+ messages in thread
* Re: lockd grace period
@ 2005-02-01 10:46 mehta kiran
  0 siblings, 0 replies; 13+ messages in thread
From: mehta kiran @ 2005-02-01 10:46 UTC (permalink / raw)
  To: nfs; +Cc: steved, chetan

Hi  , 
        I verified that suse NFS server enters
grace period whenever nfsd starts , irrespective
of whether lockd was already runnnig or not.
While Red Hat NFS server (RHEL 4 pre_rc2) does
not restart grace period of already running
lockd , even if nfsd restarts.

**Can anybody let me know what the standard behaviour
  should be. ??

What i think is that suse does it right by restarting
nfs grace period of already running lockd.
This is because machine acting as nfs server
can definitely act as nfs client of some other
machine(though rare but occurs).

Will nfs developers from Suse "also" like to comment
on this?

thanks,
-- kiran

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com 


-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl
_______________________________________________
NFS maillist  -  NFS@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/nfs

^ permalink raw reply	[flat|nested] 13+ messages in thread

end of thread, other threads:[~2005-02-01 10:46 UTC | newest]

Thread overview: 13+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-01-31  7:41 lockd grace period mehta kiran
2005-01-31 12:33 ` Steve Dickson
  -- strict thread matches above, loose matches on Subject: below --
2005-01-31 12:50 Aswathanarayana, Chetan (STSD)
2005-01-31 14:36 ` mehta kiran
2005-01-31 15:01   ` Steve Dickson
2005-01-31 15:09     ` mehta kiran
2005-01-31 16:01       ` Steve Dickson
2005-01-31 14:58 ` Steve Dickson
2005-01-31 15:01   ` mehta kiran
2005-01-31 23:07   ` Dan Stromberg
2005-01-31 23:57     ` Trond Myklebust
2005-02-01  3:07     ` Steve Dickson
2005-02-01 10:46 mehta kiran

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.