[OpenAFS] "Constant" disk writing
John W Higgins
wishdev@gmail.com
Mon, 10 Aug 2009 09:04:19 -0700
--0016e6475c7212826e0470cbbd07
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Morning Derrick,
On Mon, Aug 10, 2009 at 8:57 AM, Derrick Brashear <shadow@gmail.com> wrote:
> On Mon, Aug 10, 2009 at 11:43 AM, John W Higgins<wishdev@gmail.com> wrote:
> > Good Morning Jeffrey,
> >
> > On Mon, Aug 10, 2009 at 3:53 AM, Jeffrey Altman
> > <jaltman@secure-endpoints.com> wrote:
> >>
> >> I'm just wondering. Why do you believe that the file server
> >> (or other) processes are writing to your volumes as opposed to
> >> reading from them or writing to log (or other) files?
> >>
> >
> > First, thanks for the response.....
> >
> > I'm basing this on a couple of things - first when I shut down the
> OpenAFS
> > server the disk activity stops. Second, I ran iotop to examine the io
> usage
> > and the writes were going to the 2 loop devices I was using to test
> OpenAFS
> > and nothing else is using that area but OpenAFS.
>
> I assume iotop lists process names; Given that there's no process
> called OpenAFS, what did it really tell you?
> Does it list fds?
The process was called fileserver and it was the
/usr/libexec/openafs/fileserver (I think that's the right location) process
that was making the writes.
>
>
> > And finally, top shows
> > fileserver popping up every 10 seconds when the disk activity to the loop
> > devices is occurring and then it goes back to sleep.
> >
> > Is it possible that because I'm using loop devices something out of the
> norm
> > is occuring?
>
> Unlikely.
>
> What version of OpenAFS? I suppose it's possible the background sync
> thread is in some way hurting you.
>
I'm running 1.4.9 (which I know it's the most current so I'm more then happy
to upgrade if that appears to be the issue).
John
--0016e6475c7212826e0470cbbd07
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Morning Derrick,<br><br><div class=3D"gmail_quote">On Mon, Aug 10, 2009 at =
8:57 AM, Derrick Brashear <span dir=3D"ltr"><<a href=3D"mailto:shadow@gm=
ail.com">shadow@gmail.com</a>></span> wrote:<br><blockquote class=3D"gma=
il_quote" style=3D"border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0=
pt 0pt 0.8ex; padding-left: 1ex;">
<div class=3D"im">On Mon, Aug 10, 2009 at 11:43 AM, John W Higgins<<a hr=
ef=3D"mailto:wishdev@gmail.com">wishdev@gmail.com</a>> wrote:<br>
> Good Morning Jeffrey,<br>
><br>
> On Mon, Aug 10, 2009 at 3:53 AM, Jeffrey Altman<br>
> <<a href=3D"mailto:jaltman@secure-endpoints.com">jaltman@secure-end=
points.com</a>> wrote:<br>
>><br>
>> I'm just wondering. =A0Why do you believe that the file server=
<br>
>> (or other) processes are writing to your volumes as opposed to<br>
>> reading from them or writing to log (or other) files?<br>
>><br>
><br>
> First, thanks for the response.....<br>
><br>
> I'm basing this on a couple of things - first when I shut down the=
OpenAFS<br>
> server the disk activity stops. Second, I ran iotop to examine the io =
usage<br>
> and the writes were going to the 2 loop devices I was using to test Op=
enAFS<br>
> and nothing else is using that area but OpenAFS.<br>
<br>
</div>I assume iotop lists process names; Given that there's no process=
<br>
called OpenAFS, what did it really tell you?<br>
Does it list fds?</blockquote><div><br>The process was called fileserver an=
d it was the /usr/libexec/openafs/fileserver (I think that's the right =
location) process that was making the writes.<br>=A0<br></div><blockquote c=
lass=3D"gmail_quote" style=3D"border-left: 1px solid rgb(204, 204, 204); ma=
rgin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
<div class=3D"im"><br>
> And finally, top shows<br>
> fileserver popping up every 10 seconds when the disk activity to the l=
oop<br>
> devices is occurring and then it goes back to sleep.<br>
><br>
> Is it possible that because I'm using loop devices something out o=
f the norm<br>
> is occuring?<br>
<br>
</div>Unlikely.<br>
<br>
What version of OpenAFS? I suppose it's possible the background sync<br=
>
thread is in some way hurting you.<br>
</blockquote><div><br>I'm running 1.4.9 (which I know it's the most=
current so I'm more then happy to upgrade if that appears to be the is=
sue).<br><br>John</div></div>
--0016e6475c7212826e0470cbbd07--