Hi All,
I run the command:
[ceph@ceph-client ~]$ sudo rbd map ceph_block_dev --pool rbd --id admin -k
/etc/ceph/ceph.client.admin.keyring
ERROR: modinfo: could not find module rbd
FATAL: Module rbd not found.
rbd: modprobe rbd failed! (256)
As you may already see, there got failed!
Hello,
On Fri, 09 May 2014 08:10:54 +0200 Michal Pazdera wrote:
> Hi everyone,
>
> I am new to the Ceph. I have 5 PC test cluster on wich id like to test
> CephFS behavior and performance. I have used ceph-deploy on nod pc1 and
> installed ceph software (emeperor 0.72.2-0.el6) on all 5 machines
Hi Irek,
the default value of replication level in version firefly is 3, while in
version emperor is 2, this is reason to make my cluster unstable.
I have another issue:
the size of the folder omap of some osds is very big, it is at about 2GB -
8GB, is there any way to clean up this folder?
Best
Right,
I've run into the situation where the system seems reluctant to
reorganise after changing all the pool sizes - until the osds are
restarted (essentially I just rebooted each host in turn) *then* the
health went to OK. This was a while ago (pre 0.72), so something else
may be going on w
http://www.sebastien-han.fr/blog/2014/05/01/vagrant-up-install-ceph-in-one-command/
Sébastien Han
Cloud Engineer
"Always give 100%. Unless you're giving blood.”
Phone: +33 (0)1 49 70 99 72
Mail: sebastien@enovance.com
Address : 11 bis, rue Roquépine - 75008 Paris
Web : www.enovanc
Hi,
I experienced exactly the same with 14.04 and the 0.79 release.
It was a fresh clean install with default crushmap and ceph-deploy install
as pr. the quick-start guide.
Oddly enough changing replica size (incl min_size) from 3 - 2 (and 2->1)
and back again it worked.
I didn't have time to l
Let's assume a test cluster up and running with real data on it.
Which is the best way to migrate everything to a production (and
larger) cluster?
I'm thinking to add production MONs to the test cluster, after that,
add productions OSDs to the test cluster, waiting for a full rebalance
and then st
An other thought, I would hope that with EC, data chunks spreads would profits
of each drives writes capability where there will be stored.
I did not get any rely for now ! Does this kind of configuration (hard & soft)
looks crazy ?! Am I missing something ?
Looking forward for your comments, t
No particular reason actually. Just thought it would be simpler. However, iSCSI
looks simple enough from the howtos.
Thanks for your suggestions and I will give it a shot
- Original Message -
From: "Stuart Longland"
To: ceph-users@lists.ceph.com
Sent: Friday, 9 May, 2014 12:26:17
Ideally I would like to have a setup with 2+ iscsi servers, so that I can
perform maintenance if necessary without shutting down the vms running on the
servers. I guess multipathing is what I need.
Also I will need to have more than one xenserver/vmware host servers, so the
iscsi LUNs will be
What you need is tgtd and this patch from the ceph site
http://ceph.com/dev-notes/updates-to-ceph-tgt-iscsi-support/ for tgt-admin,
then you can set up udev rules for persistent naming on the initiator and in
turn set up multipathd – the caveat here is that we’ve never had any luck
getting more
On Fri, 9 May 2014, Indra Pramana wrote:
> Hi Gandalf and Sage,
>
> Just would like to confirm if my steps below to replace a journal disk are
> correct? Presuming the journal disk to be replaced is /dev/sdg and the two
> affected OSDs using the disk as journals are osd.30 and osd.31:
>
> - ceph
2014-05-09 15:55 GMT+02:00 Sage Weil :
> This looks correct to me!
Some command to automate this in ceph would be nice.
For example, skipping the "mkjournal" step:
ceph-osd -i 30 --mkjournal
ceph-osd -i 31 --mkjournal
ceph should be smarth enough to automatically make journals if missing
so that
I'm working on http://tracker.ceph.com/issues/8310 , basically by
bringing osds down and up I've come to a state where on-disk I have pgs,
osds seem to scan the directories on boot, but the crush map isn't
mapping the objects properly.
In addition to that ticket, I've got a decompile of my cru
So we were attempting to stress test a cephfs installation, and last night,
after copying 500GB of files, we got this:
570G in the "raw" directory
q782657@usadc-seaxd01:/mounts/ceph1/pubdata/tcga$ ls -lh
total 32M
-rw-rw-r-- 1 q783775 pipeline 32M May 8 10:39
2014-02-25T12:00:01-0800_data_man
Hi Erik,
What happens if you try to stat one of the "missing" files (assuming you know
the name of the file before you remount raw)?
I had a problem where files would disappear and reappear in CephFS, which I
believe was fixed in kernel 3.12.
Cheers,
Lincoln
On May 9, 2014, at 9:30 AM, Arones
Hi,
I'm seeing really low latency values, to the extent that they don't seem
realistic.
Snippet from the latest perf dump for this OSD:
"op_r_latency": { "avgcount": 184229,
"sum": 178.07771},
Long run avg = 178.07771/184229 = 0.00097 ms? Is it correct that latency values
have m
On Fri, 9 May 2014, Jeff Bachtel wrote:
> I'm working on http://tracker.ceph.com/issues/8310 , basically by bringing
> osds down and up I've come to a state where on-disk I have pgs, osds seem to
> scan the directories on boot, but the crush map isn't mapping the objects
> properly.
>
> In additio
178/184229=0.00097 s = 0.97ms
On Fri, May 9, 2014 at 10:49 PM, Dan Ryder (daryder) wrote:
> Hi,
>
>
>
> I’m seeing really low latency values, to the extent that they don’t seem
> realistic.
>
>
>
> Snippet from the latest perf dump for this OSD:
>
>
>
> "op_r_latency": { "avgcount": 184229,
>
>
Re-added ML.
On Fri, 9 May 2014 14:50:54 + Carlos M. Perez wrote:
> Christian,
>
> Thanks for the responses. See below for a few
> reposnses/comments/further questions...
>
> > -Original Message-
> > From: Christian Balzer [mailto:ch...@gol.com]
> > Sent: Friday, May 9, 2014 1:35
If I stat on that box, I get nothing:
q782657@usadc-seaxd01:/mounts/ceph1/pubdata/tcga/raw$ cd BRCA
-bash: cd: BRCA: No such file or directory
perl -e 'print stat("BRCA")'
If I access a mount on another machine, I can see the files:
q782657@usadc-nasea05:/mounts/ceph1/pubdata/tcga$ ls -l raw
t
I can always remount and see them.
But I wanted to preserve the "broken" state and see if I could figure out why
it was happening. (strace isn't particularly revealing.)
Some other things I noted was that
- if I reboot the metadata server nobody seems to "fail over" to the hot spare
(eve
Thanks Haomai,
So are all latency values calculated in seconds?
Dan
-Original Message-
From: Haomai Wang [mailto:haomaiw...@gmail.com]
Sent: Friday, May 09, 2014 11:20 AM
To: Dan Ryder (daryder)
Cc: ceph-us...@ceph.com
Subject: Re: [ceph-users] Low latency values
178/184229=0.00097 s =
yes
On Sat, May 10, 2014 at 12:19 AM, Dan Ryder (daryder) wrote:
> Thanks Haomai,
>
> So are all latency values calculated in seconds?
>
> Dan
>
> -Original Message-
> From: Haomai Wang [mailto:haomaiw...@gmail.com]
> Sent: Friday, May 09, 2014 11:20 AM
> To: Dan Ryder (daryder)
> Cc: cep
Ok, that makes sense for the OSD IO latency values. But I'm confused about the
recoverystate_perf latency values.
For example:
"started_latency": { "avgcount": 296,
"sum": 86047405.517876000},
"primary_latency": { "avgcount": 240,
"sum": 53489945.22253},
If these values a
I don't think anybody's done this before, but that will functionally
work, yes. Depending on how much of the data in the cluster you
actually care about, you might be better off just taking it out (rbd
export/import or something) instead of trying to incrementally move
all the data over, but...*shr
+1 with 0.80 on Ubuntu 14.04. I have a 7-node cluster with 10 OSDs per node on
a 10Gbps network. 3 of the nodes are also acting as monitors. Each node has a
single 6-core CPU, 32GB of memory, 1 SSD for journals, and 36 3TB hard drives.
I'm currently using 11 of the hard drives in each node, one
I'm less current on the kernel client, so maybe there are some
since-fixed bugs I'm forgetting, but:
On Fri, May 9, 2014 at 8:55 AM, Aronesty, Erik
wrote:
> I can always remount and see them.
>
> But I wanted to preserve the "broken" state and see if I could figure out why
> it was happening.
The recovery_state "latencies" are all about how long your PGs are in
various states of recovery; they're not per-operation latencies. 3
days still seems awfully long, but if you had a lot of data that
needed to get recovered and were throttling it tightly enough that
could happen.
-Greg
Software E
> Let's assume a test cluster up and running with real data on it.
> Which is the best way to migrate everything to a production (and
> larger) cluster?
>
> I'm thinking to add production MONs to the test cluster, after that,
> add productions OSDs to the test cluster, waiting for a full rebalance
Thanks Greg,
That makes sense.
Can you also confirm that latency values are always in seconds?
I haven't seen any documentation for it and want to be sure before I say it is
one way or the other.
Dan
-Original Message-
From: Gregory Farnum [mailto:g...@inktank.com]
Sent: Friday, May
FWIW, I believe the particular/similar bug I was thinking of was fixed by:
commit 590fb51f1c (vfs: call d_op->d_prune() before unhashing dentry)
--Lincoln
On May 9, 2014, at 12:37 PM, Gregory Farnum wrote:
> I'm less current on the kernel client, so maybe there are some
> since-fixed bu
On Fri, May 9, 2014 at 10:49 AM, Dan Ryder (daryder) wrote:
> Thanks Greg,
>
> That makes sense.
>
> Can you also confirm that latency values are always in seconds?
> I haven't seen any documentation for it and want to be sure before I say it
> is one way or the other.
I believe that's the case,
Dne 9.5.2014 9:08, Christian Balzer napsal(a):
Is that really just one disk?
Yes, its just one disk in all PCs. I know that the setup is bad, but I
want just to get
familiar with Ceph (and other parallel fs like Gluster ot Lustre) and
see what they can
do and cannot.
You have the reason for t
Sorry about not sending to list. Most of my other lists default to from the
list, not the individual...
Thanks for the links below. Equipment should be here next week, so we'll get
to do some testing...
Carlos M. Perez
CMP Consulting Services
305-669-1515
> -Original Message-
> From:
Wow I'm an idiot for getting the wrong reweight command.
Thanks so much,
Jeff
On May 9, 2014 11:06 AM, "Sage Weil" wrote:
> On Fri, 9 May 2014, Jeff Bachtel wrote:
> > I'm working on http://tracker.ceph.com/issues/8310 , basically by
> bringing
> > osds down and up I've come to a state where on
Andrey,
In initial testing, it looks like it may work rather efficiently.
1) Upgrade all mon, osd, and clients to Firefly. Restart everything so
no legacy ceph code is running.
2) Add "mon osd allow primary affinity = true" to ceph.conf, distribute
ceph.conf to nodes.
3) Inject it into t
I'm still a noob too, so don't take anything I say with much weight. I
was hoping that somebody with more experience would reply.
I see a few potential problems.
With that CPU to disk ratio, you're going to need to slow recovery down
a lot to make sure you have enough CPU available after a n
On Fri, 09 May 2014 23:03:50 +0200 Michal Pazdera wrote:
> Dne 9.5.2014 9:08, Christian Balzer napsal(a):
> > Is that really just one disk?
>
> Yes, its just one disk in all PCs. I know that the setup is bad, but I
> want just to get
> familiar with Ceph (and other parallel fs like Gluster ot Lu
39 matches
Mail list logo