On 02/10/2016 06:07 PM, Jason Dillaman wrote:
> Can you provide the 'rbd info' dump from one of these corrupt images?
>
sure,
rbd image 'ldap01.root.borked':
size 2 MB in 5000 objects
order 22 (4096 kB objects)
block_name_prefix: rbd_data.18394b3d1b58ba
The Supermicro 5018A-AR12L is built for object storage. In our testing,
they perform pretty well. You would have to invest in discrete 10G nics to
meet all of your requirements.
Austin
On Thu, Feb 11, 2016 at 8:04 PM, Schlacta, Christ
wrote:
> If you can swing 2u chassis
On Thu, Feb 11, 2016 at 4:29 PM, Blade Doyle wrote:
>
> After several months of use without needing any administration at all, I
> think I finally found something to debug.
>
> Attempting to "ls -l" within a directory on CephFS hangs - strace shows its
> hanging on lstat():
Hi,
ceph VERY SLOW with 24 osd(SAMSUNG ssd).
fio /dev/rbd0 iodepth=1 direct=1 IOPS only ~200
fio /dev/rbd0 iodepth=32 direct=1 IOPS only ~3000
But test single ssd deive with fio:
fio iodepth=1 direct=1 IOPS ~15000
fio iodepth=32 direct=1 IOPS ~3
Why ceph SO SLOW? Could you give me
Hi.
You need to read :
https://www.sebastien-han.fr/blog/2014/10/10/ceph-how-to-test-if-your-ssd-is-suitable-as-a-journal-device/
С уважением, Фасихов Ирек Нургаязович
Моб.: +79229045757
2016-02-12 10:41 GMT+03:00 Huan Zhang :
> Hi,
>
> ceph VERY SLOW with 24
Michael,
I wouldn't be concerned with SAS expanders - so long as you've got enough
bandwidth to the HBA / RAID controller?
The main consideration with the SSDs is the ratio to disk. When you loose an
SSD all the OSDs journalling to that will be inconsistent, effectively
off-lining them. Too
Thanks for all the help on this!
Regards,
Mavis
On Wed, Feb 10, 2016 at 6:00 PM, Bill Sanders
wrote:
> Going in a tiny bit more detail to what Michał said, one of the key
> reasons for having the journal (in particular, to use SSD's) is to
> reduce latency on writes
I think I see the problem. It looks like you are performing ops directly
against the cache tier instead of the base tier (assuming cache1 is your cache
pool). Here are my steps against master where the object is successfully
promoted upon 'rbd info':
# ceph osd erasure-code-profile set
That class method flag to force promotion was to prevent proxying lock/unlock
class ops. What was happening was that the cache tier OSD would become the
owner of the lock instead of the client (retrieved via cls_get_request_origin).
--
Jason Dillaman
- Original Message -
> From:
That PDF specifically calls for P3700 NVMe SSDs, not the consumer 750. You need
high endurance drives usually.
I’m using 1x400GB Intel P3700 per 9 OSDs (so 4xP3700 per 36 disk chassis).
> On 11 Feb 2016, at 17:56, Michael wrote:
>
> Alex Leake writes:
>
I'm trying to reproduce this.
Jason, I found your commit marks certain cls ops and requiring promotion,
but that doesn't include rbd... and I'm not sure why info would need to be
promoted. Working on reproducing this under hammer with the appropriate
recency settings.
sage
On Thu, 11 Feb
Hi Jason,
It's in writeback mode.
Nick
> -Original Message-
> From: ceph-devel-ow...@vger.kernel.org [mailto:ceph-devel-
> ow...@vger.kernel.org] On Behalf Of Jason Dillaman
> Sent: 11 February 2016 18:40
> To: Nick Fisk
> Cc: Sage Weil ; Samuel Just
I saw this go by in the commit log:
commit cc2200c5e60caecf7931e546f6522b2ba364227f
Merge: f8d5807 12c083e
Author: Sage Weil
Date: Thu Feb 11 08:44:35 2016 -0500
Merge pull request #7537 from ifed01/wip-no-promote-for-delete-fix
osd: fix unnecessary object
Hi Mihai, Grüezi Ivan :)
Thank both of you for the fast reply. Its appreciated.
When I bootstrapped the cluster I used
--
osd_pool_default_size = 3
osd_pool_default_min_size = 2
--
in ceph.conf. This is also set for each pool at the moment.
I understood from docs this means each object is
A "mid-cycle summit" for OpenStack operators will be held in Manchester
(England) on Monday/Tuesday next week (15/16 February).
The morning session on Tuesday will include a slot on Ceph integration.
If there are any Ceph+OpenStack operators, please have a look at the
Etherpad with the draft
Hello Michael,
I maintain a small Ceph cluster at the University of Bath, our cluster consists
of:
Monitors:
3 x Dell PowerEdge R630
- 2x Intel(R) Xeon(R) CPU E5-2609 v3
- 64GB RAM
- 4x 300GB SAS (RAID 10)
OSD Nodes:
6 x Dell PowerEdge R730XD & MD1400 Shelves
- 2x Intel(R) Xeon(R) CPU
Assuming the partition table is still zeroed on that image, can you run:
# rados -p get rbd_data.18394b3d1b58ba. - | cut -b
512 | hexdump
Can you also provide your pool setup:
# ceph report --format xml 2>/dev/null | xmlstarlet sel -t -c "//osdmap/pools"
--
Jason Dillaman
Hi All,
Thought I would just share this as it's taken me a couple of hours to work
it out.
I'm using Diamond+Carbon+Graphite to collect the perf dump output from the
admin sockets. The Diamond collector is really good at capturing everything,
but stores the sum+avgcount directly into Carbon.
In
That’s a relief, I was sensing a major case of face palm occuring when I read
Jason's email!!!
> -Original Message-
> From: ceph-devel-ow...@vger.kernel.org [mailto:ceph-devel-
> ow...@vger.kernel.org] On Behalf Of Sage Weil
> Sent: 11 February 2016 21:00
> To: Jason Dillaman
I to all,
someone have try to cancel rbd default pool?
I have the cluster ceph in warning with stale create pg.
Is possible cancel default rbd pool and remove all stale pg?
Thanks to all for your reply.
Andrea.
___
ceph-users mailing list
All,
I have a set of hardware with a few systems connected via IB along with a DDN
SFA12K.
There are 4 IB/SRP paths to each block device. Those show up as
/dev/mapper/mpath[b-d]
I am trying to do an initial install/setup of ceph on 3 nodes. Each will be a
monitor as well as host a single OSD.
After several months of use without needing any administration at all, I
think I finally found something to debug.
Attempting to "ls -l" within a directory on CephFS hangs - strace shows its
hanging on lstat():
open("/etc/group", O_RDONLY|O_CLOEXEC) = 4
fstat(4, {st_mode=S_IFREG|0644,
On Thu, 11 Feb 2016, Nick Fisk wrote:
> That’s a relief, I was sensing a major case of face palm occuring when I
> read Jason's email!!!
https://github.com/ceph/ceph/pull/7617
The tangled logic in maybe_handle_cache wasn't respecting the force
promotion bool.
Hello Andrea,
The question is why won't your PG's go into an active+clean state on the
cluster? Are all of your OSD's up/in? Are you satisfying your CRUSH ruleset?
Can you provide an output of 'ceph osd tree', 'ceph -s', 'ceph osd crush
show-tunables' and your ceph.conf file.
Thank you,
I was able to reproduce this on master:
On Thu, 11 Feb 2016, Jason Dillaman wrote:
> I think I see the problem. It looks like you are performing ops directly
> against the cache tier instead of the base tier (assuming cache1 is your
> cache pool). Here are my steps against master where the
What does 'ceph osd tree' look like for this cluster? Also have you done
anything special to your CRUSH rules?
I've usually found this to be caused by modifying OSD weights a little too
much.
As for the inconsistent PG, you should be able to run 'ceph pg repair' on
it:
Looks like it to me, yeah. Not sure why it took so long to get noticed
though (that is, is that behavior present in the release you're using,
or is it a new bug)?
-Greg
On Thu, Feb 11, 2016 at 12:11 PM, Stephen Lord wrote:
>
> I saw this go by in the commit log:
>
>
Hi All,
Has anyone seen or heard about any Xeon-D based barebones servers more
suited for Ceph use. We are looking to roll out a 2nd cluster and would
really like something based on these boards that is barebones and dense.
There is a lot of fat in most servers that we just don't need and
Hi Micheal,
ceph -s:
cluster ea296c34-e9b0-4a53-a740-f0b472f0c81d
health HEALTH_WARN
44 pgs degraded
64 pgs stale
44 pgs stuck degraded
64 pgs stuck inactive
64 pgs stuck stale
128 pgs stuck unclean
All,
Ok, it was indeed me.
Firewalld does not seem happy across boots when network manager is involved
unless you use something like nm-connection-editor to put the nic in the zone
you want... grrr
Brian Andrus
From: ceph-users [mailto:ceph-users-boun...@lists.ceph.com] On Behalf Of
Hi!
We are on version 9.2.0, 5 mons and 80 OSDS distributed on 10 hosts.
How could we twist cephx capabilities so to forbid our KVM+QEMU+libvirt
hosts any RBD creation capability ?
We currently have an rbd-user key like so :
caps: [mon] allow r
caps: [osd] allow x
If you can swing 2u chassis and 2.5" drives instead, you can trivially get
between 15 and 24 drives across the front and rear of a beautiful hot-swap
chassis. There are numerous makes and models available from custom builds
up/down through used on ebay. Worth a peek.
On Thu, Feb 11, 2016 at
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256
Is this only a problem with EC base tiers or would replicated base
tiers see this too?
-
Robert LeBlanc
PGP Fingerprint 79A2 9CA4 6CC4 45DD A904 C70E E654 3BB2 FA62 B9F1
On Thu, Feb 11, 2016 at 6:09 PM, Sage Weil wrote:
> On
Hello,
As discussed on this very ML and by looking at the moderately well hidden
full specifications, using any 7xx Intel SSD for journals is a fools
errand.
Their endurance is far too low in comparison to their IOPS capacity.
Regards,
Chibi
On Thu, 11 Feb 2016 15:15:12 + Alex Leake
Alex Leake writes:
>
> Hello Michael,
>
> I maintain a small Ceph cluster at the University of Bath, our cluster
consists of:
>
> Monitors:
> 3 x Dell PowerEdge R630
>
> - 2x Intel(R) Xeon(R) CPU E5-2609 v3
> - 64GB RAM
> - 4x 300GB SAS (RAID 10)
>
> OSD Nodes:
> 6 x
Hi,
Are there any tips and tricks around getting rid of misplaced objects? I did
check the archive but didn’t find anything.
Right now my cluster looks like this:
pgmap v43288593: 16384 pgs, 4 pools, 45439 GB data, 10383 kobjects
109 TB used, 349 TB / 458 TB avail
Hi Sage,
Do you think this will get fixed in time for the Jewel release? It still seems
to happen in Master and is definitely related to the recency setting. I'm
guessing that the info command does some sort of read and then a write. In the
old behaviour the read would have always triggered a
What's your cache mode? In the master branch, I would expect that class method
ops should force a promotion to the cache tier if the base tier is an EC pool
[1].
[1] https://github.com/ceph/ceph/blob/master/src/osd/OSD.cc#L8905
--
Jason Dillaman
- Original Message -
> From: "Nick
38 matches
Mail list logo