Hi,
We have a 3-site Ceph cluster and would like to create a 4+2 EC pool
with 2 chunks per datacenter, to maximise the resilience in case of 1
datacenter being down. I have not found a way to create an EC profile
with this 2-level allocation strategy. I created an EC profile with a
failure do
===
Frank Schilder
AIT Risø Campus
Bygning 109, rum S14
____
From: Michel Jouvin
Sent: Monday, April 3, 2023 6:40 PM
To: ceph-users@ceph.io
Subject: [ceph-users] Crushmap rule for multi-datacenter erasure coding
Hi,
We have a 3-site Ceph cluster and would like t
without me doing overtime on incidents. Its much cheaper in the long run. A
50-60% usable-capacity cluster is very easy and cheap to administrate.
Best regards,
=
Frank Schilder
AIT Risø Campus
Bygning 109, rum S14
________
From: Michel Jouvin
Sent: Mon
Hi,
As discussed in another thread (Crushmap rule for multi-datacenter
erasure coding), I'm trying to create an EC pool spanning 3 datacenters
(datacenters are present in the crushmap), with the objective to be
resilient to 1 DC down, at least keeping the readonly access to the pool
and if po
, l=4
configuration is equivalent, in terms of redundancy, to a jerasure
configuration with k=9, m=6.
Michel
Le 04/04/2023 à 15:26, Michel Jouvin a écrit :
Hi,
As discussed in another thread (Crushmap rule for multi-datacenter
erasure coding), I'm trying to create an EC pool spanning 3
Hi,
We have a 3-site Ceph cluster and would like to create a 4+2 EC pool
with 2 chunks per datacenter, to maximise the resilience in case of 1
datacenter being down. I have not found a way to create an EC profile
with this 2-level allocation strategy. I created an EC profile with a
failure do
.
Thanks in advance if somebody could provide some sort of authoritative
answer on these 2 questions. Best regards,
Michel
Le 04/04/2023 à 15:53, Michel Jouvin a écrit :
Answering to myself, I found the reason for 2147483647: it's
documented as a failure to find enough OSD (missing OSDs). And
Hi,
Our cluster is running Pacific 16.2.10. We have a problem using the
dashboard to display information about RGWs configured in the cluster.
When clicking on "Object Gateway", we get an error 500. Looking in the
mgr logs, I found that the problem is that the RGW is accessed by its IP
addres
rsion 16.2.11 (which was just recently released) contains a fix for
that. But it still doesn’t work with wildcard certificates, that’s
still an issue for us.
Zitat von Michel Jouvin :
Hi,
Our cluster is running Pacific 16.2.10. We have a problem using the
dashboard to display information about
Hi,
For what is worth, we have a similar problem in 16..2.10 that I had no time
to troubleshoot yet. It happened after adding a haproxy in front of rgw to
manage https and switch rgw to http (to overcome the other pb mentioned
when using https in rgw). The access/secret key is refused despite
Hi Marc,
I can share what we did a few months ago. As a remark, I am not sure
Nautilus is available in EL8 but may be I missed it. In our case we did
the following travel:
- Pacific to Octopus on EL7, traditionally managed
- Conversion of the cluster to a cephadm cluster as it makes every
up
Hi,
I just upgraded in 17.2.6 but in fact I had the same problem in 16.2.10.
I'm trying to configure the Ceph dashboard to monitor the RGWs (object
gateways used as S3 gw). Our cluster has 2 RGW realms (eros, fink) with
1 zonegroup per realm (p2io-eros and p2io-fink respectively) and 1 zone
p
à 12:55, Michel Jouvin a écrit :
Hi,
I just upgraded in 17.2.6 but in fact I had the same problem in
16.2.10. I'm trying to configure the Ceph dashboard to monitor the
RGWs (object gateways used as S3 gw). Our cluster has 2 RGW realms
(eros, fink) with 1 zonegroup per realm (p2io-eros and
down (5 for R/W access) so I'm wondering if there is something wrong in
our pool configuration (k=9, m=6, l=5).
Cheers,
Michel
Le 06/04/2023 à 08:51, Michel Jouvin a écrit :
Hi,
Is somebody using LRC plugin ?
I came to the conclusion that LRC k=9, m=3, l=4 is not the same as
jerasure
Hi Wesley,
I can only answer your second question and give an opinion on the last one!
- Yes the OSD activation problem (in cephadm clusters only) was
introduce by an unfortunate change (indentation problem in Python code)
in 16.2.11. The issue doesn't exist in 16.2.10 and is one of the fixed
that as
I was using the LRC plugin, I had the warranty that I could loose a site
without impact, thus the possibility to loose 1 OSD server. Am I wrong?
Best regards,
Michel
Le 24/04/2023 à 13:24, Michel Jouvin a écrit :
Hi,
I'm still interesting by getting feedback from those using
t a écrit :
Hello,
What is your current setup, 1 server pet data center with 12 osd each?
What is your current crush rule and LRC crush rule?
On Fri, Apr 28, 2023, 12:29 Michel Jouvin
wrote:
Hi,
I think I found a possible cause of my PG down but still
understand why.
As explai
SD), min_size 4. But as I wrote, it probably doesn't
have the resiliency for a DC failure, so that needs some further
investigation.
Regards,
Eugen
Zitat von Michel Jouvin :
Hi,
No... our current setup is 3 datacenters with the same
configuration, i.e. 1 mon/mgr + 4 OSD servers with 16 OSDs
ck a écrit :
Hi,
I don't think you've shared your osd tree yet, could you do that?
Apparently nobody else but us reads this thread or nobody reading this
uses the LRC plugin. ;-)
Thanks,
Eugen
Zitat von Michel Jouvin :
Hi,
I had to restart one of my OSD server today and the prob
e only
way I can currently visualize it working is with more servers, I'm thinking
6 or 9 per data center min, but that could be my lack of knowledge on some
of the step rules.
Thanks
Curt
On Tue, May 16, 2023 at 11:09 AM Michel Jouvin <
michel.jou...@ijclab.in2p3.fr> wrote:
Hi
interested.
Best regards,
Michel
Le 21/05/2023 à 16:07, Michel Jouvin a écrit :
Hi Eugen,
My LRC pool is also somewhat experimental so nothing really urgent. If
you manage to do some tests that help me to understand the problem I
remain interested. I propose to keep this thread for that
Hi Patrick,
It is weird, we have a couple of clusters with cephadm and running pacify
or quincy and ceph orch device works well. Have you looked at the cephadm
logs (ceph log last cephadm)?
Except if you are using a very specific hardware, I suspect Ceph is
suffering of a problem outside it.
e to Pacific?
Could this be a workaround for this sort of regression from Octopus to
Pacific ?
May be updating the BIOS from 1.7.1 to 1.8.1 ?
All this is a little bit confusing for me as I'm trying to discover Ceph 😁
Thanks
Patrick
Le 26/05/2023 à 17:19, Michel Jouvin a écrit :
Hi Patrick
+1
Michel
Le 30/05/2023 à 11:23, Frank Schilder a écrit :
What I'm having in mind is if the command is already in history. A wrong history
reference can execute a command with "--yes-i-really-mean-it" even though you
really don't mean it. Been there. For an OSD this is maybe tolerable, but fo
Hi David,
Normally cephadm connection issue are not that difficult to solve. It is
just the matter of having the appropriate SSH configuration in the root
account. Mainly the public key used by cephadm (extracted with the
command you used in a shell) added in the root account .ssh/authorized_k
ove to have a comment from the developers on this topic. But in
this state I would not recommend to use the LRC plugin when the
resiliency requirements are to sustain the loss of an entire DC.
Thanks,
Eugen
[1]
https://docs.ceph.com/en/latest/rados/operations/erasure-code-lrc/#low-level-plu
,
Michel
Le 26/05/2023 à 18:50, Michel Jouvin a écrit :
Patrick,
I can only say that I would not expect a specific problem due to your
hardware. Upgrading the firmware is generally a good idea but I
wouldn't expect it helps in your case if the osk (lsblk) sees the disk.
As for starting
Hi Niklas,
I am not sure why you are surprised. In a large cluster, you should
expect some rebalancing on every crush map or crush map rule change.
Ceph doesn't just enforce the failure domain, it also whants to have a
"perfect" pseudo-random distribution across the clusters based on the
crus
Hi Niklas,
As I said, ceph placement is based on more than fulfilling the failure
domain constraint. This is a core feature in ceph design. There is no
reason for a rebalancing on a cluster with a few hundreds OSDs to last a
month. Just before 17 you have to adjust the max backfills parameter
Hi Herve,
Why you don't use the automatic adjustment of the number of PGs. This
makes life much easier and works well.
Cheers,
Michel
Le 03/10/2023 à 17:06, Hervé Ballans a écrit :
Hi all,
Sorry for the reminder, but does anyone have any advice on how to deal
with this?
Many thanks!
Her
Hi,
I'm trying to use the rgw mgr module to configure RGWs. Unfortunately it
is not present in 'ceph mgr module ls' list and any attempt to enable it
suggests that one mgr doesn't support it and that --force should be
added. Adding --force effectively enabled it.
It is strange as it is a bra
Hi,
I'm struggling with a problem to add cephadm some hosts in our Quincy
cluster. "ceph orch host add host addr" fails with the famous "missing 2
required positional arguments: 'hostname' and 'addr'" because of bug
https://tracker.ceph.com/issues/59081 but looking at cephadm messages
with "c
1 Gb management network so no point to use
Jumbo frames). I cannot think of anything that Ceph could have mentioned
to help diagnose this.
Best regards,
Michel
Le 25/10/2023 à 14:42, Michel Jouvin a écrit :
Hi,
I'm struggling with a problem to add cephadm some hosts in our Quincy
cluster.
Hi Albert,
You should never edit any file in the containers, cephadm takes care of
it. Most of the parameters described in the doc you mentioned are better
managed with "ceph config" command in the Ceph configuration database.
If you want to run the ceph commnand on a Ceph machine outside a
c
Hi,
We have a Reef cluster that started to complain a couple of weeks ago
about ~20 PGs (over 10K) not scrubbed/deep-scrubbed in time. Looking at
it since a few days, I saw this affect only those PGs that could not be
scrubbed since mid-February. Old the other PGs are regularly scrubbed.
I d
Hi Rafael,
Good to know I am not alone!
Additional information ~6h after the OSD restart: over the 20 PGs
impacted, 2 have been processed successfully... I don't have a clear
picture on how Ceph prioritize the scrub of one PG over another, I had
thought that the oldest/expired scrubs are take
low
Universitätsrechenzentrum (URZ)
Universität Greifswald
Felix-Hausdorff-Straße 18
17489 Greifswald
Germany
Tel.: +49 3834 420 1450
--- Original Nachricht ---
*Betreff: *[ceph-users] Re: Reef (18.2): Some PG not scrubbed/deep
scrubbed for 1 month
*Von: *"Michel Jouvin" <mailto:michel.j
still flooded with "scrub starts" and i have no clue
why these OSDs are causing the problems.
Will investigate further.
Best regards,
Gunnar
===
Gunnar Bandelow
Universitätsrechenzentrum (URZ)
Universität Greifswald
Felix-Hausdorff-Straße 18
17489 Gr
onfiguration setting if you
haven't already, but note that it can impact client I/O performance.
Also, if the delays appear to be related to a single OSD, have you
checked the health and performance of this device?
On Fri, 22 Mar 2024 at 09:29, Michel Jouvin
wrote:
Hi,
As I s
en setting osd_op_queue back to mpq for
this only OSD would probably reveal it. Not sure about the implication of
having a signel OSD running a different scheduler in the cluster though.
- Le 22 Mar 24, à 10:11, Michel Jouvin michel.jou...@ijclab.in2p3.fr a
écrit :
Pierre,
Yes, as ment
rvers at some points. It is unclear why
it is not enough to just rerun the benchmark and why a crazy value for
an HDD is found...
Best regards,
Michel
Le 22/03/2024 à 14:44, Michel Jouvin a écrit :
Hi Frédéric,
I think you raise the right point, sorry if I misunderstood Pierre's
suggesti
OSD would probably reveal it. Not sure
about the implication of having a signel OSD running a different
scheduler in the cluster though.
>
>
> - Le 22 Mar 24, à 10:11, Michel Jouvin
michel.jou...@ijclab.in2p3.fr a écrit :
>
>> Pierre,
>>
ing osd_op_queue
back to mpq for this only OSD would probably reveal it. Not sure
about the implication of having a signel OSD running a different
scheduler in the cluster though.
>
>
> - Le 22 Mar 24, à 10:11, Michel Jouvin
michel.jou...@ijclab.in2p3.fr a écri
Nima,
Can you also specify the Ceph version you are using and whether your
current configuration is cephadm-based?
Michel
Le 13/05/2024 à 15:19, Götz Reinicke a écrit :
Hi,
Am 11.05.2024 um 15:54 schrieb Nima AbolhassanBeigi
:
Hi,
We want to upgrade the OS version of our production ceph
Hi,
We have a cephadm-based Octopus (upgraded to 15.2.17 today but the
problem started with 15.2.16) cluster where we try to deploy a RGW in
multisite configuration. We followed the documentation at
https://docs.ceph.com/en/octopus/radosgw/multisite/ to do the basic
realm, zonegroup, zone and
One additional information that may be relevant for the problem: the
server hosting the RGW has 2 networks configured on the same interface
and the default one is not the Ceph public network but another network
not related to Ceph. I suspect a Podman configuration issue where the
default networ
Hi,
Unfortunately it was a wrong track. The problem remains the same, with
the same error messages, on another host with only one network address
in the Ceph cluster public network. BTW, "ceph shell --name rgw_daemon"
works and from the shell I can use radosgw-admin and ceph command,
suggesti
ter starting) but it
makes diagnostics of such trivial misconfiguration difficults...
Cheers,
Michel
Le 29/09/2022 à 14:04, Michel Jouvin a écrit :
Hi,
Unfortunately it was a wrong track. The problem remains the same, with
the same error messages, on another host with only one network address
Hi,
We have a production cluster made of 12 OSD servers with 16 OSD each
(all the same HW) which has been running fine for 5 years (initially
installed with Luminous) and which has been running Octopus (15.2.16)
for 1 year and was recently upgraded to 15.2.17 (1 week before the
problem starte
CSI bus when failing. Stopping
the osd revives the rest of the disks on the box).
Cheers, Dan
On Sun, Oct 16, 2022, 22:08 Michel Jouvin wrote:
Hi,
We have a production cluster made of 12 OSD servers with 16 OSD each
(all the same HW) which has been running fine for 5 years
received. Cheers,
Michel
Le 16/10/2022 à 22:49, Michel Jouvin a écrit :
Hi,
In fact, a very stupid mistake. This is a CentOS 8 system where smartd
was not installed. After installing and starting it, the OSD device is
indeed in bad shape... Sorry for the noise.
Cheers,
Michel
Le 16/10/2022 à 22
Hi,
Eugen is right, no need to drain and readd your OSDs, this will be very
long and lead to a lot of unnecessary load balancing. You should
decouple replacing the failed OSD (where you should try to drain it,
after setting its primary-affinity to 0 to limit the I/O redirected to
it, and read
52 matches
Mail list logo