vstart.sh crashes MON with --paxos-propose-interval=0.01 and one MDS

[Loic Dachary]

Hi John,

I would be gratefull if you could take a quick look at http://tracker.ceph.com/issues/9794 . It is bisected to the reduction of pg and I'm able to reproduce it in a ubuntu-14.04 docker fresh install. For some reason it does not happen in gitbuilder but I think you can reproduce it locally now.

Cheers

[David Zafman]

I have this change in my branch so that test/ceph_objectstore_tool.py w=
orks again after that change from John. I wonder if this would fix you=
r case too:

commit 18937cf49be616d32b4e2d0b6deef2882321fbe4
Author: David Zafman <***@redhat.com>
Date: Tue Oct 14 18:45:41 2014 -0700

vstart.sh: Disable "mon pg warn min per osd" to get healthy

Signed-off-by: David Zafman <***@redhat.com>

diff --git a/src/vstart.sh b/src/vstart.sh
index febfa56..7a0ec1c 100755
--- a/src/vstart.sh
+++ b/src/vstart.sh
@@ -394,7 +394,7 @@ $COSDDEBUG
$COSDMEMSTORE
$extra_conf
[mon]
- mon pg warn min per osd =3D 10
+ mon pg warn min per osd =3D 0
mon osd allow primary affinity =3D true
mon reweight min pgs per osd =3D 4
$DAEMONOPTS

David Zafman
Senior Developer
http://www.inktank.com

Post by Loic Dachary
Hi John,
=20
I would be gratefull if you could take a quick look at http://tracker=

=2Eceph.com/issues/9794 . It is bisected to the reduction of pg and I'm=
able to reproduce it in a ubuntu-14.04 docker fresh install. For some =
reason it does not happen in gitbuilder but I think you can reproduce i=
t locally now.

Post by Loic Dachary
=20
Cheers
=20
--=20
Lo=EFc Dachary, Artisan Logiciel Libre
=20

--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" i=
n
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html

[Joao Eduardo Luis]

I have this change in my branch so that test/ceph_objectstore_tool.py=

works again after that change from John. I wonder if this would fix y=

commit 18937cf49be616d32b4e2d0b6deef2882321fbe4
Date: Tue Oct 14 18:45:41 2014 -0700
vstart.sh: Disable "mon pg warn min per osd" to get healthy
diff --git a/src/vstart.sh b/src/vstart.sh
index febfa56..7a0ec1c 100755
--- a/src/vstart.sh
+++ b/src/vstart.sh
@@ -394,7 +394,7 @@ $COSDDEBUG
$COSDMEMSTORE
$extra_conf
[mon]
- mon pg warn min per osd =3D 10
+ mon pg warn min per osd =3D 0
mon osd allow primary affinity =3D true
mon reweight min pgs per osd =3D 4
$DAEMONOPTS

Even if that fixes the issue you're seeing, that is nothing but a=20
band-aid. I've tracked down the issue Loic is seeing a few days ago an=
d=20
it is sitting in wip-9321.giant=20
(c84a13ae87eed5550bafda394d983a8e843cc08c). I though it was due to an=20
earlier patch of mine on the same branch but now I see the issue can po=
p=20
up regardless of my patch, which is a good thing as it as it doesn't=20
mean we really have to wait until that branch clears up the suite to=20
merge but simply current master with this patch applied.

I've updated #9794 with this information.

-Joao

David Zafman
Senior Developer
http://www.inktank.com

Post by Loic Dachary
Hi John,
I would be gratefull if you could take a quick look at http://tracke=

r.ceph.com/issues/9794 . It is bisected to the reduction of pg and I'm =
able to reproduce it in a ubuntu-14.04 docker fresh install. For some r=
eason it does not happen in gitbuilder but I think you can reproduce it=
locally now.

Post by Loic Dachary
Cheers
--
Lo=EFc Dachary, Artisan Logiciel Libre

--
To unsubscribe from this list: send the line "unsubscribe ceph-devel"=

in

More majordomo info at http://vger.kernel.org/majordomo-info.html

--=20
Joao Eduardo Luis
Software Engineer | http://inktank.com | http://ceph.com
--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" i=
n
the body of a message to ***@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html