1 Dec
2020
1 Dec
'20
2:55 p.m.
I've been working on a patch series to overhaul the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated
filesystem mounted on /var/cache/fscache, so that if it fills up it
doesn't take down the rootfs with it. We'll also need to have
cachefilesd installed and running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
Thanks,
--
Jeff Layton <jlayton(a)redhat.com>
1 Dec
1 Dec
3:22 p.m.
On 12/1/20 9:55 AM, Jeff Layton wrote:
I've been working on a patch series to overhaul
the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated
filesystem mounted on /var/cache/fscache, so that if it fills up it
doesn't take down the rootfs with it. We'll also need to have
cachefilesd installed and running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
I think I can make this happen pretty easily in ceph-cm-ansible. What
I'd need from you is desired filesystem type and size. Once I'm done
with my end, we'll need to create a cephlab.ansible overrides yaml
fragment to stick in that suite's qa directory.
3:24 p.m.
On Tue, 2020-12-01 at 10:22 -0500, David Galloway wrote:
On 12/1/20 9:55 AM, Jeff Layton wrote:
Ok, cool:
fstype: xfs or ext4 (either is fine)
size: ~50g or so would be ideal, but we can probably get away with less
if necessary
Thanks!
--
Jeff Layton <jlayton(a)redhat.com>
I've been working on a patch series to
overhaul the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated
filesystem mounted on /var/cache/fscache, so that if it fills up it
doesn't take down the rootfs with it. We'll also need to have
cachefilesd installed and running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
I think I can make this happen pretty easily in ceph-cm-ansible. What
I'd need from you is desired filesystem type and size. Once I'm done
with my end, we'll need to create a cephlab.ansible overrides yaml
fragment to stick in that suite's qa directory.
7:23 p.m.
On 12/1/20 10:24 AM, Jeff Layton wrote:
On Tue, 2020-12-01 at 10:22 -0500, David Galloway
wrote:
Getting there...
https://github.com/ceph/ceph-cm-ansible/pull/592
Can you give me an optimal cachefilesd.conf and I'll set that up in
ceph-cm-ansible too?
On 12/1/20 9:55 AM, Jeff Layton wrote:
Ok, cool:
fstype: xfs or ext4 (either is fine)
size: ~50g or so would be ideal, but we can probably get away with less
if necessary
I've been working on a patch series to
overhaul the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated
filesystem mounted on /var/cache/fscache, so that if it fills up it
doesn't take down the rootfs with it. We'll also need to have
cachefilesd installed and running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
I think I can make this happen pretty easily in ceph-cm-ansible. What
I'd need from you is desired filesystem type and size. Once I'm done
with my end, we'll need to create a cephlab.ansible overrides yaml
fragment to stick in that suite's qa directory.
8:07 p.m.
On Tue, 2020-12-01 at 14:23 -0500, David Galloway wrote:
On 12/1/20 10:24 AM, Jeff Layton wrote:
I never actually tweak the defaults in my local testing, so let's start
with a default cachefilesd.conf as installed by the package. If we need
to tweak it later, then we'll go from there.
Thanks!
--
Jeff Layton <jlayton(a)redhat.com>
On Tue, 2020-12-01 at 10:22 -0500, David Galloway
wrote:
Getting there...
https://github.com/ceph/ceph-cm-ansible/pull/592
Can you give me an optimal cachefilesd.conf and I'll set that up in
ceph-cm-ansible too?
On 12/1/20 9:55 AM, Jeff Layton wrote:
Ok, cool:
fstype: xfs or ext4 (either is fine)
size: ~50g or so would be ideal, but we can probably get away with less
if necessary
I've been working on a patch series to
overhaul the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated
filesystem mounted on /var/cache/fscache, so that if it fills up it
doesn't take down the rootfs with it. We'll also need to have
cachefilesd installed and running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
I think I can make this happen pretty easily in ceph-cm-ansible. What
I'd need from you is desired filesystem type and size. Once I'm done
with my end, we'll need to create a cephlab.ansible overrides yaml
fragment to stick in that suite's qa directory.
9:31 p.m.
On 12/1/20 3:07 PM, Jeff Layton wrote:
On Tue, 2020-12-01 at 14:23 -0500, David Galloway
wrote:
K. I made it optional and used the package-provided defaults.
Your requested changes are implemented in ceph-cm-ansible. I've
attached the yaml fragments you'll need to pass on the
`teuthology-suite` command line to set up the partition, filesystem, and
cachefilesd service.
e.g., teuthology-suite ... ~/fs-cache-smithi.yml
On 12/1/20 10:24 AM, Jeff Layton wrote:
I never actually tweak the defaults in my local testing, so let's start
with a default cachefilesd.conf as installed by the package. If we need
to tweak it later, then we'll go from there.
On Tue, 2020-12-01 at 10:22 -0500, David Galloway
wrote:
Getting there...
https://github.com/ceph/ceph-cm-ansible/pull/592
Can you give me an optimal cachefilesd.conf and I'll set that up in
ceph-cm-ansible too?
On 12/1/20 9:55 AM, Jeff Layton wrote:
> I've been working on a patch series to overhaul the fscache code in the
> kclient. I also have this (really old) tracker to add fscache testing to
> teuthology:
>
> https://tracker.ceph.com/issues/6373
>
> It would be ideal if the clients in such testing had a dedicated
> filesystem mounted on /var/cache/fscache, so that if it fills up it
> doesn't take down the rootfs with it. We'll also need to have
> cachefilesd installed and running in the client hosts.
>
> Is it possible to do this in teuthology? How would I approach this?
>
I think I can make this happen pretty easily in ceph-cm-ansible. What
I'd need from you is desired filesystem type and size. Once I'm done
with my end, we'll need to create a cephlab.ansible overrides yaml
fragment to stick in that suite's qa directory.
Ok, cool:
fstype: xfs or ext4 (either is fine)
size: ~50g or so would be ideal, but we can probably get away with less
if necessary
9:59 p.m.
On Tue, 2020-12-01 at 16:31 -0500, David Galloway wrote:
On 12/1/20 3:07 PM, Jeff Layton wrote:
Many thanks! I now need to figure out how to pass arbitrary mount
options to the kclient. I think that might need some plumbing down in
ceph/qa. I'll sort that out soon and test this.
Cheers!
--
Jeff Layton <jlayton(a)redhat.com>
On Tue, 2020-12-01 at 14:23 -0500, David Galloway
wrote:
K. I made it optional and used the package-provided defaults.
Your requested changes are implemented in ceph-cm-ansible. I've
attached the yaml fragments you'll need to pass on the
`teuthology-suite` command line to set up the partition, filesystem, and
cachefilesd service.
e.g., teuthology-suite ... ~/fs-cache-smithi.yml
On 12/1/20 10:24 AM, Jeff Layton wrote:
I never actually tweak the defaults in my local testing, so let's start
with a default cachefilesd.conf as installed by the package. If we need
to tweak it later, then we'll go from there.
On Tue, 2020-12-01 at 10:22 -0500, David Galloway
wrote:
> On 12/1/20 9:55 AM, Jeff Layton wrote:
> > I've been working on a patch series to overhaul the fscache code in the
> > kclient. I also have this (really old) tracker to add fscache testing to
> > teuthology:
> >
> > https://tracker.ceph.com/issues/6373
> >
> > It would be ideal if the clients in such testing had a dedicated
> > filesystem mounted on /var/cache/fscache, so that if it fills up it
> > doesn't take down the rootfs with it. We'll also need to have
> > cachefilesd installed and running in the client hosts.
> >
> > Is it possible to do this in teuthology? How would I approach this?
> >
>
> I think I can make this happen pretty easily in ceph-cm-ansible. What
> I'd need from you is desired filesystem type and size. Once I'm done
> with my end, we'll need to create a cephlab.ansible overrides yaml
> fragment to stick in that suite's qa directory.
>
Ok, cool:
fstype: xfs or ext4 (either is fine)
size: ~50g or so would be ideal, but we can probably get away with less
if necessary
Getting there...
https://github.com/ceph/ceph-cm-ansible/pull/592
Can you give me an optimal cachefilesd.conf and I'll set that up in
ceph-cm-ansible too?
2 Dec
2 Dec
2:49 a.m.
Hi Sepia and Ceph Team,
Looking for some help, wanted to configure CEPH storage (over PROXMOX) as
an IaaS for my home/apartment CCTV - NVR project to storage 40HD camera
streams, where I can find a Step by Step guide to configure and implement
the same?
Thanks,
-ashish
-----Original Message-----
From: Jeff Layton [mailto:jlayton@redhat.com]
Sent: 1 December 2020 08:25 PM
To: dev <dev(a)ceph.io>io>; Ceph Development <ceph-devel(a)vger.kernel.org>
Cc: sepia <sepia(a)ceph.io>
Subject: [sepia] provisioning clients in teuthology with an extra local
filesystem
I've been working on a patch series to overhaul the fscache code in the
kclient. I also have this (really old) tracker to add fscache testing to
teuthology:
https://tracker.ceph.com/issues/6373
It would be ideal if the clients in such testing had a dedicated filesystem
mounted on /var/cache/fscache, so that if it fills up it doesn't take down
the rootfs with it. We'll also need to have cachefilesd installed and
running in the client hosts.
Is it possible to do this in teuthology? How would I approach this?
Thanks,
--
Jeff Layton <jlayton(a)redhat.com>
_______________________________________________
Sepia mailing list -- sepia(a)ceph.io
To unsubscribe send an email to sepia-leave(a)ceph.io
1240
days inactive
1241
days old
7 comments
3 participants
participants (3)
-
Ashish -
David Galloway -
Jeff Layton