Hi,
Thanks for looking into this, this seems to mitigate the problem.
Do you think this is just related to httpd or is going to impact other
services such as nginx and be a wider point to know about / document?
Thanks
On Mon, Oct 24, 2016 at 2:28 PM, Yan, Zheng wrote:
> I
I finally reproduced this issue. Adding following lines to httpd.conf
can workaround this issue.
EnableMMAP off
EnableSendfile off
On Sat, Sep 3, 2016 at 11:07 AM, Yan, Zheng wrote:
> On Fri, Sep 2, 2016 at 5:10 PM, Sean Redmond wrote:
>> I have
On Fri, Sep 2, 2016 at 5:10 PM, Sean Redmond wrote:
> I have checked all the servers in scope running 'dmesg | grep -i stale' and
> it does not yield any results.
>
> As a test I have rebooted the servers in scope and I can still replicate the
> behavior 100% of the time.
Hi,
The work around is fine, I still think there could be a bug here. I will
try and spend some time in the next few days to write something to test
with Regular buffered IO.
But even using memory-mapped IO I would not expect the read request from
server1 or server2 to get zeros's in place of
On Fri, Sep 2, 2016 at 11:35 AM, Sean Redmond wrote:
> Hi,
>
> That makes sense, I have worked around this by forcing the sync within the
> application running under apache and it is working very well now without the
> need for the 'sync' mount option.
>
> What
Hi,
That makes sense, I have worked around this by forcing the sync within the
application running under apache and it is working very well now without
the need for the 'sync' mount option.
What interesting is that in the pastebin provided below it shows a way to
replicate this, I was just using
On Thu, Sep 1, 2016 at 8:02 AM, Sean Redmond wrote:
> Hi,
>
> It seems to be using syscall mmap() from what I read this indicates it is
> using memory-mapped IO.
>
> Please see a strace here: http://pastebin.com/6wjhSNrP
Zheng meant is Apache using memory-mapped IO. From
I have checked all the servers in scope running 'dmesg | grep -i stale' and
it does not yield any results.
As a test I have rebooted the servers in scope and I can still replicate
the behavior 100% of the time.
On Fri, Sep 2, 2016 at 4:37 AM, Yan, Zheng wrote:
> I think
I think about this again. This issue could be caused by stale session.
Could you check kernel logs of your servers. Are there any ceph
related kernel message (such as "ceph: mds0 caps stale")
Regards
Yan, Zheng
On Thu, Sep 1, 2016 at 11:02 PM, Sean Redmond wrote:
> Hi,
Hi,
It seems to be using syscall mmap() from what I read this indicates it is
using memory-mapped IO.
Please see a strace here: http://pastebin.com/6wjhSNrP
Thanks
On Wed, Aug 31, 2016 at 5:51 PM, Sean Redmond
wrote:
> I am not sure how to tell?
>
> Server1 and
I am not sure how to tell?
Server1 and Server2 mount the ceph file system using kernel client 4.7.2
and I can replicate the problem using '/usr/bin/sum' to read the file or a
http GET request via a web server (apache).
On Wed, Aug 31, 2016 at 2:38 PM, Yan, Zheng wrote:
> On
On Wed, Aug 31, 2016 at 12:49 AM, Sean Redmond wrote:
> Hi,
>
> I have been able to pick through the process a little further and replicate
> it via the command line. The flow seems looks like this:
>
> 1) The user uploads an image to webserver server 'uploader01' it gets
It seems using the 'sync' mount option on the server uploader01 is also a
valid work around.
Is it a problem that the meta data is available to other cephfs clients
ahead of the file contents being flushed by the client doing the write?
I think having an invalid page cache of zeros is a problem
Hi,
I have been able to pick through the process a little further and replicate
it via the command line. The flow seems looks like this:
1) The user uploads an image to webserver server 'uploader01' it gets
written to a path such as '/cephfs/webdata/static/456/JHL/66448H-755h.jpg'
on cephfs
2)
You are correct it only seems to impact recently modified files.
On Tue, Aug 30, 2016 at 3:36 AM, Yan, Zheng wrote:
> On Tue, Aug 30, 2016 at 2:11 AM, Gregory Farnum
> wrote:
> > On Mon, Aug 29, 2016 at 7:14 AM, Sean Redmond
>
On Tue, Aug 30, 2016 at 2:11 AM, Gregory Farnum wrote:
> On Mon, Aug 29, 2016 at 7:14 AM, Sean Redmond wrote:
>> Hi,
>>
>> I am running cephfs (10.2.2) with kernel 4.7.0-1. I have noticed that
>> frequently static files are showing empty when serviced
Hi,
Yes the file has no contents until the page cache is flushed.
I will give the fuse client a try and report back.
Thanks
On Mon, Aug 29, 2016 at 7:11 PM, Gregory Farnum wrote:
> On Mon, Aug 29, 2016 at 7:14 AM, Sean Redmond
> wrote:
> > Hi,
>
On Mon, Aug 29, 2016 at 7:14 AM, Sean Redmond wrote:
> Hi,
>
> I am running cephfs (10.2.2) with kernel 4.7.0-1. I have noticed that
> frequently static files are showing empty when serviced via a web server
> (apache). I have tracked this down further and can see when
Hi,
I am running cephfs (10.2.2) with kernel 4.7.0-1. I have noticed that
frequently static files are showing empty when serviced via a web server
(apache). I have tracked this down further and can see when running a
checksum against the file on the cephfs file system on the node serving the
Hi,
I am running cephfs (10.2.2) with kernel 4.7.0-1. I have noticed that
frequently static files are showing empty when serviced via a web server
(apache). I have tracked this down further and can see when running a
checksum against the file on the cephfs file system on the node serving the
20 matches
Mail list logo