[OpenAFS] Re: kernel panic in 1.4.14.1 on rhel6
Aaron Knister
aaronk@umbc.edu
Mon, 15 Aug 2011 18:55:48 -0400
--bcaec54ee9422cfd0a04aa932bd0
Content-Type: text/plain; charset=ISO-8859-1
On Mon, Aug 15, 2011 at 6:02 PM, Derrick Brashear <shadow@gmail.com> wrote:
> On Mon, Aug 15, 2011 at 4:26 PM, Aaron Knister <aaronk@umbc.edu> wrote:
> > On Fri, Aug 12, 2011 at 12:01 PM, Andrew Deason <adeason@sinenomine.net>
> > wrote:
> >>
> >> On Thu, 11 Aug 2011 10:11:27 -0400
> >> Aaron Knister <aaronk@umbc.edu> wrote:
> >>
> >> > I know everyone's busy, but out of curiosity has anybody had a chance
> >> > to look at this? Could it be that it's not even an AFS bug, but a
> >> > linux kernel issue? I'm getting ready to put a webserver into
> >> > production and am concerned because I can reliably trigger the
> >> > mentioned kernel panic using "ab".
> >>
> >> I don't think it's likely that it's a linux kernel issue. If you want
> >> something to try, you can try applying these two patches:
> >>
> >>
> >>
> http://git.openafs.org/?p=openafs.git;a=commitdiff_plain;h=910f3d4cbba4cd78a8e6afa0b32e37ed7f292d18
> >>
> >>
> http://git.openafs.org/?p=openafs.git;a=commitdiff_plain;h=35835af9bb85b358e4bdd11113b637ac0e52c56b
> >>
> >> Those fix what may be a similar issue, but I'm not really sure if it's
> >> the same thing as here. If that doesn't do it (it's good that you have a
> >> reliable way of reproducing this), we'll need to look more closely at
> >> the core (if you don't want to provide it since it probably contains
> >> file data, instructions can probably be provided, though it's a lot more
> >> detailed than what we've done so far). Or we can proably go the route of
> >> giving you debug patches, but I'm not quite sure what to add quite yet.
> >>
> >
> > Hi Andrew,
> > Those patches do prevent it from crashing, but also results in a number
> of
> > "permission denied" errors from the apache processes under heavy load
> from
> > ab. The permission denied errors don't occur with 1.6.0pre7.
>
> is there a reason 1.6.0pre7 isn't suitable, incidentally?
>
Not really, other than some of my colleagues had reservations about using a
pre-release in production.
>
> --
> Derrick
>
--
Aaron Knister
Systems Administrator
Division of Information Technology
University of Maryland, Baltimore County
aaronk@umbc.edu
--bcaec54ee9422cfd0a04aa932bd0
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
<div class=3D"gmail_quote">On Mon, Aug 15, 2011 at 6:02 PM, Derrick Brashea=
r <span dir=3D"ltr"><<a href=3D"mailto:shadow@gmail.com">shadow@gmail.co=
m</a>></span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"margi=
n:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
<div><div></div><div class=3D"h5">On Mon, Aug 15, 2011 at 4:26 PM, Aaron Kn=
ister <<a href=3D"mailto:aaronk@umbc.edu">aaronk@umbc.edu</a>> wrote:=
<br>
> On Fri, Aug 12, 2011 at 12:01 PM, Andrew Deason <<a href=3D"mailto:=
adeason@sinenomine.net">adeason@sinenomine.net</a>><br>
> wrote:<br>
>><br>
>> On Thu, 11 Aug 2011 10:11:27 -0400<br>
>> Aaron Knister <<a href=3D"mailto:aaronk@umbc.edu">aaronk@umbc.e=
du</a>> wrote:<br>
>><br>
>> > I know everyone's busy, but out of curiosity has anybody =
had a chance<br>
>> > to look at this? Could it be that it's not even an AFS bu=
g, but a<br>
>> > linux kernel issue? I'm getting ready to put a webserver =
into<br>
>> > production and am concerned because I can reliably trigger th=
e<br>
>> > mentioned kernel panic using "ab".<br>
>><br>
>> I don't think it's likely that it's a linux kernel iss=
ue. If you want<br>
>> something to try, you can try applying these two patches:<br>
>><br>
>><br>
>> <a href=3D"http://git.openafs.org/?p=3Dopenafs.git;a=3Dcommitdiff_=
plain;h=3D910f3d4cbba4cd78a8e6afa0b32e37ed7f292d18" target=3D"_blank">http:=
//git.openafs.org/?p=3Dopenafs.git;a=3Dcommitdiff_plain;h=3D910f3d4cbba4cd7=
8a8e6afa0b32e37ed7f292d18</a><br>
>><br>
>> <a href=3D"http://git.openafs.org/?p=3Dopenafs.git;a=3Dcommitdiff_=
plain;h=3D35835af9bb85b358e4bdd11113b637ac0e52c56b" target=3D"_blank">http:=
//git.openafs.org/?p=3Dopenafs.git;a=3Dcommitdiff_plain;h=3D35835af9bb85b35=
8e4bdd11113b637ac0e52c56b</a><br>
>><br>
>> Those fix what may be a similar issue, but I'm not really sure=
if it's<br>
>> the same thing as here. If that doesn't do it (it's good t=
hat you have a<br>
>> reliable way of reproducing this), we'll need to look more clo=
sely at<br>
>> the core (if you don't want to provide it since it probably co=
ntains<br>
>> file data, instructions can probably be provided, though it's =
a lot more<br>
>> detailed than what we've done so far). Or we can proably go th=
e route of<br>
>> giving you debug patches, but I'm not quite sure what to add q=
uite yet.<br>
>><br>
><br>
> Hi Andrew,<br>
> Those patches do prevent it from crashing, but also results in a numbe=
r of<br>
> "permission denied" errors from the apache processes under h=
eavy load from<br>
> ab. The permission denied errors don't occur with 1.6.0pre7.<br>
<br>
</div></div>is there a reason 1.6.0pre7 isn't suitable, incidentally?<b=
r></blockquote><div><br></div><div>Not really, other than some of my collea=
gues had reservations about using a pre-release in production.=A0</div><div=
>
=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;borde=
r-left:1px #ccc solid;padding-left:1ex;">
<br>
--<br>
<font color=3D"#888888">Derrick<br>
</font></blockquote></div><br><br clear=3D"all"><div><br></div>-- <br>Aaron=
Knister<br>Systems Administrator<br>Division of Information Technology<br>=
University of Maryland, Baltimore County<br><a href=3D"mailto:aaronk@umbc.e=
du" target=3D"_blank">aaronk@umbc.edu</a><br>
--bcaec54ee9422cfd0a04aa932bd0--