Re: nfsd (and lock) changes for 3.2

[Boaz Harrosh <bharrosh@panasas.com>]

On 10/27/2011 01:34 PM, J. Bruce Fields wrote:
> On Tue, Oct 25, 2011 at 08:16:08PM -0700, Boaz Harrosh wrote:
>> On 10/25/2011 05:19 AM, J. Bruce Fields wrote:
>>> Boaz Harrosh (1):
>>>       nfsd4: fix failure to end nfsd4 grace period
>>>
>>
>> Bruce hi.
>>
>> As I recall this failure I have experienced in early stages of the
>> 3.1-rcX series. I admit that I always run with Benny's pnfs tree,
>> but I do not think it is patches that Benny added from the 3.2 set
>> I think it is breakage introduced by the 3.1 merge window.
>>
>> Therefor I think that this patch must be sent to Stable@ otherwise
>> surly 3.1 Kernel users are to hit it. (You can escapee it)
> 
> We need to test it, but I think you're probably correct,
> 6577aac01f00636c16cd583c30bd4dedf18475d5 "nfsd4: fix failure to end
> nfsd4 grace period" should have been tagged for stable.  Does that also
> require any previous patches? (E.g.
> 48483bf23a568f3ef4cc7ad2c8f1a082f10ad0e7 "nfsd4: simplify recovery dir
> setting"?)
> 

right, and all for that dmesg print. But yes! 

Well the 3.1 Kernel without them gets stuck. I guess you need to experiment
with what would be the minimal patchset needed for 3.1 .

Tell me if you need help testing that the damage has gone. But basically I have
a script (4 years old) that does: 
	server: service nfs stop/start 
	client: mount
back to back, The client gets stuck forever.

> --b.

Thanks
Boaz