denied
3/26/24 9:38:09 PM[INF]Updating
dkcphhpcmgt028:/var/lib/ceph/5c384430-da91-11ed-af9c-c780a5227aff/config/ceph.conf
It seem to be related to the permissions that the manager writes the files with
and the process copying them around.
$ sudo ceph -v
[sudo] password for adminjskr:
ceph version 17.2.6 (d7ff0d1
more resources on recovery than 328 MiB/s
Thanks, .
--
Jesper Krogh
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
pg_num and pgp_num took immidiate effect?
Jesper
jskr@dkcphhpcmgt028:/$ sudo ceph version
ceph version 17.2.6 (d7ff0d10654d2280e08f1ab989c7cdf3064446a5) quincy
(stable)
jskr@dkcphhpcmgt028:/$ sudo ceph health
HEALTH_OK
jskr@dkcphhpcmgt028:/$ sudo ceph status
cluster:
id: 5c384430-da91
Hi.
I have a Ceph (NVME) based cluster with 12 hosts and 40 OSD's .. currently it
is backfilling pg's but I cannot get it to run more than 20 backfilling (pgs)
at the same time (6+2 profile)
osd_max_backfills = 100 and osd_recovery_max_active_ssd = 50 (non-sane) but it
still stops at 20 with
p OSD - repair - mark-complete
on the primary OSD.
A scrub tells me that the "active+clean” state is for real.
I also found out the more automated "force-create-pg" command only works on pgs
that a in down state.
Best,
Jesper
------
Jesper Lykkegaar
y with
something like this?:
* set nobackfill and norecovery
* delete the pgs shards one by one
* unset nobackfill and norecovery
Any idea on how to proceed from here is most welcome.
Thanks,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
?
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Universitetsbyen 81
8000 Aarhus C
E-mail: je...@mbg.au.dk
Tlf:+45 50906203
> On 1 Sep 2022, at 22.01, Jes
it came back online.
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Universitetsbyen 81
8000 Aarhus C
E-mail: je...@mbg.au.dk
Tlf:+45 50906203
> On 31 Aug 2
objects in that PG (also with
objectstore-tool), but this process is extremely slow.
When looping over the >65,000 object, each remove takes ~10 sec and is very
compute intensive, which is approximately 7.5 days.
Is the a faster way to get around this?
Mvh. Jes
Actually, it might have worked better if the PG had stayed down while running
cephfs-data-scan, as it could only then get file structure from metadata pool
and not touch each file/link in data pool?
This would at least properly have given the list of files in (only) the
affected PG?
//Jesper
Fra: Patrick Donnelly
Sendt: 19. august 2022 16:16
Til: Jesper Lykkegaard Karlsen
Cc: ceph-users@ceph.io
Emne: Re: [ceph-users] Potential bug in cephfs-data-scan?
On Fri, Aug 19, 2022 at 5:02 AM Jesper Lykkegaard Karlsen
wrote:
>>
> >Hi,
recently deprecated Octopus, I suspect that this bug
is also present in Pacific and Quincy?
It might be related to this bug?
https://tracker.ceph.com/issues/46166
But symptoms are different.
Or, maybe there is a way to disable the following of symlinks in
"cephfs-data-scan pg_
Cool thanks a lot!
I will definitely put it in my toolbox.
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Universitetsbyen 81
8000 Aarhus C
E-mail: je
the same amount of PGs are
backfilling.
Can large disk usage on mons slow down backfill and other operations?
Is it dangerous?
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Hi Frank,
I guess there is alway the possibility to set quota on pool level with
"target_max_objects" and “target_max_bytes”
The cephfs quotas through attributes are only for sub-directories as far as I
recall.
Best,
Jesper
------
Jesper Lykkegaard Karlsen
Ah I see, should have look at the “raw” data instead ;-)
Then I agree this very weird?
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Universitetsbyen 81
8000
as two OSDs down.
Also, I believe that min_size should never be smaller than “coding” shards,
which is 4 in this case.
You can either make a new test setup with your three test OSD hosts using EC
2+1 or make e.g. 4+2, but with failure domain set to OSD.
Best,
Jesper
or not?
Summer vacation?
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Universitetsbyen 81
8000 Aarhus C
E-mail: je...@mbg.au.dk
Tlf:+45 50906203
Thanks for you answer Janne.
Yes, I am also running "ceph osd reweight" on the "nearfull" osds, once they
get too close for comfort.
But I just though a continuous prioritization of rebalancing PGs, could make
this process more smooth, with less/no need for handheld opera
if the priority of backfill_wait PGs
was rerun, perharps every 24 hours, as OSD availability landscape of course
changes during backfill.
I imagine this, especially, could stabilize recovery/rebalance on systems where
space is a little tight.
Best regards,
Jesper
--
Jespe
dir.rbytes $i 2>/dev/nul) | sed -r 's/([0-9])([a-zA-Z])/\1 \2/g;
s/([a-zA-Z])([0-9])/\1 \2/g') $i"
fi
done
The above can be run as:
ceph_du_dir $DIR
with multiple directories:
ceph_du_dir $DIR1 $DIR2 $DIR3 ..
Or even with wildcard:
ceph_du_dir $DIR/*
Best,
Jesper
----
Not to spam, but to make it output prettier, one can also separate the number
from the byte-size prefix.
numfmt --to=iec --suffix=B --padding=7 $(getfattr --only-values -n
ceph.dir.rbytes $1 2>/dev/nul) | sed -r 's/([0-9])([a-zA-Z])/\1 \2/g;
s/([a-zA-Z])([0-9])/\1 \2/g'
//Jes
Brilliant, thanks Jean-François
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Gustav Wieds Vej 10
8000 Aarhus C
E-mail: je...@mbg.au.dk
Tlf:+45 50906203
in "human-readble"
It works like a charm and my god it is fast!.
Tools like that could be very useful, if provided by the development team
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular
Woops, wrong copy/pasta:
getfattr -n ceph.dir.rbytes $DIR
works on all distributions I have tested.
It is:
getfattr -d -m 'ceph.*' $DIR
that does not work on Rocky Linux 8, Ubuntu 18.04, but works on CentOS 7.
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific
Just tested:
getfattr -n ceph.dir.rbytes $DIR
Works on CentOS 7, but not on Ubuntu 18.04 eighter.
Weird?
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Gustav
no output. Should that not list all atributes?
This is on Rocky Linux kernel 4.18.0-348.2.1.el8_5.x86_64
Best,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Gustav Wieds
,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus University
Gustav Wieds Vej 10
8000 Aarhus C
E-mail: je...@mbg.au.dk
Tlf:+45 50906203
some "ceph snapshot recover" command, that can move metadata
pointers and make recovery much lighter, or is this just that way it is?
Best reagards,
Jesper
------
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecul
?
Anyway, I am just asking for your opinion on this? Pros and cons or even better
somebody who actually have tried this?
Best regards,
Jesper
--
Jesper Lykkegaard Karlsen
Scientific Computing
Centre for Structural Biology
Department of Molecular Biology and Genetics
Aarhus
Can i do that - when the SSDs are allready used in another crush rule -
backing and kvm_ssd rbd’s?
Jesper
Sent from myMail for iOS
Saturday, 12 September 2020, 11.01 +0200 from anthony.da...@gmail.com
:
>If you have capacity to have both online at the same time, why not add the
&g
> I would like to change the crush rule so data lands on ssd instead of hdd,
> can this be done on the fly and migration will just happen or do I need to
> do something to move data?
I would actually like to relocate my object store to a new storage tier.
Is the best to:
1) create new pool on
Hi
I would like to change the crush rule so data lands on ssd instead of hdd, can
this be done on the fly and migration will just happen or do I need to do
something to move data?
Jesper
Sent from myMail for iOS
___
ceph-users mailing list
unless uou have enabled some balancing - then this is very normal (actually
pretty good normal)
Jesper
Sent from myMail for iOS
Thursday, 14 May 2020, 09.35 +0200 from Florent B. :
>Hi,
>
>I have something strange on a Ceph Luminous cluster.
>
>All OSDs have the same size,
Hi.
How do I find out if the MDS is "busy" - being the one limiting CephFS
metadata throughput. (12.2.8).
$ time find . | wc -l
1918069
real8m43.008s
user0m2.689s
sys 0m7.818s
or 3.667ms per file.
In the light of "potentially batching" and a network latency of ~0.20ms to
the MDS -
everything again. I may get
equally good data locality with Ceph-based SSD as with local HDDs (which I
currently have)
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
is a bit more
than what looks feasible at the moment.
Thanks for your reflections/input.
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
ible of those files to other location.
* recreate pgs: ceph osd force-create-pg
* restart recovery:ceph osd unset norecover
* copy back in the recovered files
Would that work or do you have a better suggestion?
Ch
; 34400070
inodes in use by clients, 3293 stray files
Thanks - Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
iver a fast write cache for smallish writes.
Would setting the parameter til 1MB be "insane"?
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
will be as slow as hitting dead-rust - anything that
cannot live
with that need to be entirely on SSD/NVMe.
Other?
Thanks for your input.
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
Hi.
Unless there is plans for going to Petabyte scale with it - then I really
dont see the benefits of getting CephFS involved over just an RBD image
with a VM running standard samba on top.
More performant and less complexity to handle - zero gains (by my book)
Jesper
> Hi,
>
> I am
But is random/sequential read performance still good? even during saturated
write performance ?
if so the tradeoff could fit quite some applications
Sent from myMail for iOS
Friday, 6 March 2020, 14.06 +0100 from vitalif :
>Hi,
>
>Current QLC drives are total shit in terms of
Hi
We have some oldish servers with ssds - all on 25gbit nics. R815 AMD - 2,4ghz+
Is there significant performance benefits in moving to a new NVMe based, new
cpus?
+20% IOPs? + 50% IOPs?
Jesper
Sent from myMail for iOS
___
ceph-users mailing
fraction of our dataset - and 10GB cache on all 113 HDD
~1TB effective read-cache - and then writes hitting the battery-backed
write-cache - this can overspill and when hitting "cold" data performance
varies. But the read/write amplification of EC is still un-manageable in
p
fordable - or have I missed something that makes the math
work ?
--
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
h Battery Backed write-cache - will allow OSD to ack
writes before hitting spinning rust.
* More memory for OSD-level read-caching.
* 3x replication instead of EC
.. (we have all above in a "similar" setup ~1PB - 10 OSD - hosts).
SSD-tiering pool (havent been there - but would like to test it out).
--
Jesper
___
ceph-users mailing list -- ceph-users@ceph.io
To unsubscribe send an email to ceph-users-le...@ceph.io
3201013c940bf000
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3e1d0452edceebb903d23db53201013c940bf000
Was capable of deadlocking the kernel when memory pressure caused MDS to
reclaim capabilities - smells similar.
Jesper
yes
Sent from myMail for iOS
Thursday, 29 August 2019, 15.52 +0200 from fr...@dtu.dk :
>I would be in.
>
>=
>Frank Schilder
>AIT Risø Campus
>Bygning 109, rum S14
>
>
>From: Torben Hørup < tor...@t-hoerup.dk >
>Sent: 29 August 2019
concurrency is widely different 1:30
Jesper
Sent from myMail for iOS
Tuesday, 27 August 2019, 16.25 +0200 from linghucongs...@163.com
:
>The performance with the dd and fio diffrent is so huge?
>
>I have 25 OSDS with 8TB hdd. with dd I only get 410KB/s read perfomance,but
&g
What will actually happen if an old client comes by, potential data damage - or
just broken connections from the client?
jesper
Sent from myMail for iOS
Monday, 26 August 2019, 20.16 +0200 from Paul Emmerich
:
>4.13 or newer is enough for upmap
>
>--
>Paul Emmerich
>
&g
51 matches
Mail list logo