Following PRs are optimization we (Kuaishou) made for machine learning
workloads (randomly read billions of small files) .
[1] https://github.com/ceph/ceph/pull/39315
[2] https://github.com/ceph/ceph/pull/43126
[3] https://github.com/ceph/ceph/pull/43125
The first PR adds an option that disables
Hi Frank,
I think the snapshot rotation could be an explanation.
Just a few days ago we had a host failure over night and some OSDs
couldn't be rebalanced entirely because they were too full. Deleting a
few (large) snapshots I created last week resolved the issue. If you
monitored 'ceph osd
Ok, I think I found the basic problem.
I used to talk to the endpoint that is also the Domain for the s3websites.
After switching the domains around everything worked fine. :partyemote:
I have wrote down what I think how things work together (wrote down here
IYAI https://pastebin.com/6Gj9Q5hJ), an
Hi,
db_slots is still not implemented:
pacific:~ # ceph orch apply -i osd.yml --dry-run
Error EINVAL: Failed to validate Drive Group: Filtering for
is not supported
Question 2: If db_slots still *doesn't* work, is there a coherent
way to divide up a solid state DB drive for use by a bun
It happened again today:
2021-09-15 04:25:20.551098 [INF] Health check cleared: POOL_NEAR_FULL (was: 1
pools nearfull)
2021-09-15 04:19:01.512425 [INF] Health check cleared: POOL_FULL (was: 1 pools
full)
2021-09-15 04:19:01.512389 [WRN] Health check failed: 1 pools nearfull
(POOL_NEAR_FULL)
Hi,
ceph-crash services are standalone containers, they are not running
inside other containers:
host1:~ # ceph orch ls
NAME RUNNING REFRESHED AGE PLACEMENT
IMAGE NAME
IMAGE ID
Got it: one instance per host is enough.
In my case, I'm not using "ceph orch".
We did it manually, crafting one docker-compose.yml per host.
The question is:
Is it possible to run a "crash instance" per host or the solution oblige
me to adopt the cephadm solution?
Thanks!
[]'s
Arthur
On 1
Hi,
I rebooted one of my ceph nodes this morning after OS updates. No ceph
packages were upgraded. After reboot, 4 out of 12 OSDs on this host refuse
to start, giving errors:
```
Sep 15 14:59:25 janky ceph-osd[12384]: 2021-09-15T14:59:24.994+
7f418196ef00 -1 bluestore(/var/lib/ceph/osd/ceph-0
Just realised the debug paste I sent was for OSD 5 but the other info is for
OSD 0. They are both having the same issue, but for completeness sake here
is the debug output from OSD 0:
http://paste.debian.net/1211873/
All daemons in the cluster are running ceph pacific 16.2.5.
Regards,
Davíð
On W
M.2 was not designed for hot swap, and Icydock's solution is a bit outside
specification.
I really like the new Supermicro box (610P) that has 12 spinning disks and then
6 NVMs.
2 of them in 2.5"x7mm format and 4 of them in the new E1.S format.
E1.S is practically next gen hot plug M.2
Ján Senk
On 9/15/21 18:06, Davíð Steinn Geirsson wrote:
Just realised the debug paste I sent was for OSD 5 but the other info is for
OSD 0. They are both having the same issue, but for completeness sake here
is the debug output from OSD 0:
http://paste.debian.net/1211873/
All daemons in the cluster are r
Hi,
On Wed, Sep 15, 2021 at 08:39:11PM +0200, Stefan Kooman wrote:
> On 9/15/21 18:06, Davíð Steinn Geirsson wrote:
> > Just realised the debug paste I sent was for OSD 5 but the other info is for
> > OSD 0. They are both having the same issue, but for completeness sake here
> > is the debug outpu
On 9/15/21 21:02, Davíð Steinn Geirsson wrote:
Hi,
On Wed, Sep 15, 2021 at 08:39:11PM +0200, Stefan Kooman wrote:
On 9/15/21 18:06, Davíð Steinn Geirsson wrote:
Just realised the debug paste I sent was for OSD 5 but the other info is for
OSD 0. They are both having the same issue, but for comp
On Wed, Sep 15, 2021 at 09:16:17PM +0200, Stefan Kooman wrote:
> On 9/15/21 21:02, Davíð Steinn Geirsson wrote:
> > Hi,
> >
> > On Wed, Sep 15, 2021 at 08:39:11PM +0200, Stefan Kooman wrote:
> > > On 9/15/21 18:06, Davíð Steinn Geirsson wrote:
> > > > Just realised the debug paste I sent was for O
Den tors 16 sep. 2021 kl 06:28 skrev Szabo, Istvan (Agoda)
:
>
> Hi,
>
> Something weird happening, I have on 1 nvme drive and 3x SSD's are using for
> wal and db.
> The LVM is 596GB but in the health detail is says x GiB spilled over to slow
> device, however just 317 GB use only :/
>
> [WRN] BL
I haven't tried it myself but it would probably work to run the crash
services apart from cephadm, maybe someone else has a similar setup.
Since those are not critical services you can try it without impacting
the rest of the cluster. But again, I haven't tried it this way.
Zitat von Guilh
16 matches
Mail list logo