On Sep 12, 2014, at 9:09 AM, Pranith Kumar Karampuri <[email protected]> 
wrote:

> 
> On 09/11/2014 11:38 PM, mike wrote:
>> Any more to this thread? I don't mean to nag, but this seems like a pretty 
>> serious issue.
> Did you get a chance to try this test case when write-behind translator is 
> disabled?
> 
> gluster volume set <volname> performance.write-behind off

I just ran the test and it seems that disabling write-behind makes the problem 
go away.

I'll update the bug, thanks.

> 
> Pranith
>> 
>> How can I help?
>> 
>> On Sep 7, 2014, at 9:51 AM, mike <[email protected]> wrote:
>> 
>>> I don't think I have these enabled. How can I confirm that?
>>> 
>>> On Sep 7, 2014, at 12:57 AM, Anand Avati <[email protected]> wrote:
>>> 
>>>> The only reason O_APPEND gets stripped on the server side, is because of 
>>>> one of the following xlators:
>>>> 
>>>> - stripe
>>>> - quiesce
>>>> - crypt
>>>> 
>>>> If you have any of these, please try unloading/reconfiguring without these 
>>>> features and try again.
>>>> 
>>>> Thanks
>>>> 
>>>> 
>>>> On Sat, Sep 6, 2014 at 3:31 PM, mike <[email protected]> wrote:
>>>> I was able to narrow it down to smallish python script.
>>>> 
>>>> I've attached that to the bug.
>>>> 
>>>> https://bugzilla.redhat.com/show_bug.cgi?id=1138970
>>>> 
>>>> 
>>>> On Sep 6, 2014, at 1:05 PM, Justin Clift <[email protected]> wrote:
>>>> 
>>>> > Thanks Mike, this is good stuff. :)
>>>> >
>>>> > + Justin
>>>> >
>>>> >
>>>> > On 06/09/2014, at 8:19 PM, mike wrote:
>>>> >> I upgraded the client to Gluster 3.5.2, but there is no difference.
>>>> >>
>>>> >> The bug is almost certainly in the Fuse client. If I remount the 
>>>> >> filesystem with NFS, the problem is no longer observable.
>>>> >>
>>>> >> I spent a little time looking through the xlator/fuse-bridge to see 
>>>> >> where the offsets are coming from, but I'm really not familiar enough 
>>>> >> with the code, so it is slow going.
>>>> >>
>>>> >> Unfortunately, I'm still having trouble reproducing this in a python 
>>>> >> script that could be readily attached to a bug report.
>>>> >>
>>>> >> I'll take a crack at that again, but I will a file a bug anyway for 
>>>> >> completeness.
>>>> >>
>>>> >> On Sep 5, 2014, at 7:10 PM, mike <[email protected]> wrote:
>>>> >>
>>>> >>> I have narrowed down the source of the bug.
>>>> >>>
>>>> >>> Here is an strace of glusterfsd http://fpaste.org/131455/40996378/
>>>> >>>
>>>> >>> The first line represents a write that does *not* make it into the 
>>>> >>> underlying file.
>>>> >>>
>>>> >>> The last line is the write that stomps the earlier write.
>>>> >>>
>>>> >>> As I said, the client file is opened in O_APPEND mode, but on the 
>>>> >>> glusterfsd side, the file is just O_CREAT|O_WRONLY. The means the 
>>>> >>> offsets to pwrite() need to be valid.
>>>> >>>
>>>> >>> I correlated this to a tcpdump I took and I can see that in fact, the 
>>>> >>> RPCs being sent have the wrong offset.  Interestingly, 
>>>> >>> glusterfs.write-is-append = 0, which I wouldn't have expected.
>>>> >>>
>>>> >>> I think the bug lies in the glusterfs fuse client.
>>>> >>>
>>>> >>> As to your question about Gluster 3.5.2, I may be able to do that if I 
>>>> >>> am unable to find the bug in the source.
>>>> >>>
>>>> >>> -Mike
>>>> >>>
>>>> >>> On Sep 5, 2014, at 6:16 PM, Justin Clift <[email protected]> wrote:
>>>> >>>
>>>> >>>> On 06/09/2014, at 12:10 AM, mike wrote:
>>>> >>>>> I have found that the O_APPEND flag is key to this failure - I had 
>>>> >>>>> overlooked that flag when reading the strace and trying to cobble up 
>>>> >>>>> a minimal reproduction.
>>>> >>>>>
>>>> >>>>> I now have a small pair of python scripts that can reliably 
>>>> >>>>> reproduce this failure.
>>>> >>>>
>>>> >>>>
>>>> >>>> As a thought, is there a reasonable way you can test this on 
>>>> >>>> GlusterFS 3.5.2?
>>>> >>>>
>>>> >>>> There were some important bug fixes in 3.5.2 (from 3.5.1).
>>>> >>>>
>>>> >>>> Note I'm not saying yours is one of them, I'm just asking if it's
>>>> >>>> easy to test and find out. :)
>>>> >>>>
>>>> >>>> Regards and best wishes,
>>>> >>>>
>>>> >>>> Justin Clift
>>>> >>>>
>>>> >>>> --
>>>> >>>> GlusterFS - http://www.gluster.org
>>>> >>>>
>>>> >>>> An open source, distributed file system scaling to several
>>>> >>>> petabytes, and handling thousands of clients.
>>>> >>>>
>>>> >>>> My personal twitter: twitter.com/realjustinclift
>>>> >>>>
>>>> >>>
>>>> >>
>>>> >
>>>> > --
>>>> > GlusterFS - http://www.gluster.org
>>>> >
>>>> > An open source, distributed file system scaling to several
>>>> > petabytes, and handling thousands of clients.
>>>> >
>>>> > My personal twitter: twitter.com/realjustinclift
>>>> >
>>>> 
>>>> _______________________________________________
>>>> Gluster-users mailing list
>>>> [email protected]
>>>> http://supercolony.gluster.org/mailman/listinfo/gluster-users
>>>> 
>>> 
>> 
>> 
>> 
>> _______________________________________________
>> Gluster-users mailing list
>> [email protected]
>> http://supercolony.gluster.org/mailman/listinfo/gluster-users
> 
> 

_______________________________________________
Gluster-users mailing list
[email protected]
http://supercolony.gluster.org/mailman/listinfo/gluster-users

Reply via email to