hi all,
maybe to clarify a bit, e.g.
https://indico.cern.ch/event/755842/contributions/3243386/attachments/1784159/2904041/2019-jcollet-openlab.pdf
clearly shows that the db+wal disks are not saturated,
but we are wondering what is really needed/acceptable wrt throughput and
latency (eg is a
hi marc,
> - how to prevent the D state process to accumulate so much load?
you can't. in linux, uninterruptable tasks themself count as "load",
this does not mean you eg ran out of cpu resources.
stijn
>
> Thanks,
>
>
>
>
>
> ___
> ceph-users
hi,
can you give soem more details on the setup? number and size of osds.
are you using EC or not? and if so, what EC parameters?
thanks,
stijn
On 02/26/2018 08:15 AM, Linh Vu wrote:
> Sounds like you just need more RAM on your MDS. Ours have 256GB each, and the
> OSD nodes have 128GB each.
hi oliver,
>>> in preparation for production, we have run very successful tests with large
>>> sequential data,
>>> and just now a stress-test creating many small files on CephFS.
>>>
>>> We use a replicated metadata pool (4 SSDs, 4 replicas) and a data pool with
>>> 6 hosts with 32 OSDs each,
hi oliver,
the IPoIB network is not 56gb, it's probably a lot less (20gb or so).
the ib_write_bw test is verbs/rdma based. do you have iperf tests
between hosts, and if so, can you share those reuslts?
stijn
> we are just getting started with our first Ceph cluster (Luminous 12.2.2) and
>
or do it live https://access.redhat.com/articles/3311301
# echo 0 > /sys/kernel/debug/x86/pti_enabled
# echo 0 > /sys/kernel/debug/x86/ibpb_enabled
# echo 0 > /sys/kernel/debug/x86/ibrs_enabled
stijn
On 01/05/2018 12:54 PM, David wrote:
> Hi!
>
> nopti or pti=off in kernel options
the others. Plus different log sizes? It's not making a ton of sense
> at first glance.
> -Greg
>
> On Thu, Oct 19, 2017 at 1:08 AM Stijn De Weirdt <stijn.dewei...@ugent.be>
> wrote:
>
>> hi greg,
>>
>> i attached the gzip output of the query and som
any other relevant data. You shouldn't need to do manual repair of
> erasure-coded pools, since it has checksums and can tell which bits are
> bad. Following that article may not have done you any good (though I
> wouldn't expect it to hurt, either...)...
> -Greg
>
> On Wed, Oc
hi all,
we have a ceph 10.2.7 cluster with a 8+3 EC pool.
in that pool, there is a pg in inconsistent state.
we followed http://ceph.com/geen-categorie/ceph-manually-repair-object/,
however, we are unable to solve our issue.
from the primary osd logs, the reported pg had a missing object.
we
fast your processor should
> be... But making it based on how much GHz per TB is an invitation to
> context switch to death.
>
> On Sat, Aug 12, 2017, 8:40 AM Stijn De Weirdt <stijn.dewei...@ugent.be>
> wrote:
>
>> hi all,
>>
>> thanks for all the
dation will be more realistic than it was in the past (an
> effective increase in memory needs), but also that it will be under much
> better control than previously.
>
> On Thu, Aug 10, 2017 at 1:35 AM Stijn De Weirdt <stijn.dewei...@ugent.be>
> wrote:
>
>> hi all
hi all,
we are planning to purchse new OSD hardware, and we are wondering if for
upcoming luminous with bluestore OSDs, anything wrt the hardware
recommendations from
http://docs.ceph.com/docs/master/start/hardware-recommendations/
will be different, esp the memory/cpu part. i understand from
did you recreate new rpms with same version/release? it would be better
to make new rpms with different release (e.g. 9.2.0-1). we have
snapshotted mirrors and nginx caches between ceph yum repo and the nodes
that install the rpms, so cleaning the cache locally will not help.
stijn
On 11/11/2015
>>> So: the key thing to realise is that caching behaviour is full of
>>> tradeoffs, and this is really something that needs to be tunable, so
>>> that it can be adapted to the differing needs of different workloads.
>>> Having an optional "hold onto caps for N seconds after file close"
>>> sounds
hi jan,
The answer to this, as well as life, universe and everything, is simple:
ZFS.
is it really the case for ceph? i briefly looked in the filestore code a
while ago, since zfs is COW, i expected not to have a journal with ZFS,
but i couldn't find anything that suggested this was supported
Like a lot of system monitoring stuff, this is the kind of thing that
in an ideal world we wouldn't have to worry about, but the experience
in practice is that people deploy big distributed storage systems
without having really good monitoring in place. We (people providing
not to become
i would certainly like that all client libs and/or kernel modules stay
tested and supported on these OSes for future ceph releases. not sure
how much work that is, but the at least client side shouldn't be
affected by the init move.
stijn
On 07/30/2015 04:43 PM, Marc wrote:
Hi,
much like
wouldn't it be nice that ceph does something like this in background
(some sort of network-scrub). debugging network like this is not that
easy (can't expect admins to install e.g. perfsonar on all nodes and/or
clients)
something like: every X min, each service X pick a service Y on another
if you don't have 2 powerfeeds, don't spend the money.
if you have 2 feeds, well, start with 2 PSUs for your switches ;)
if you stick with one PSU for the OSDs, make sure you have your cabling
(power and network, don't forget your network switches should be on same
power feeds ;) and crushmap
hi michal,
thanks for the info. we will certainly try it and see if we come to the
same conclusions ;)
one small detail: since you were using centos7, i'm assuming you were
using ZoL 0.6.3?
stijn
On 10/29/2014 08:03 PM, Michal Kozanecki wrote:
Forgot to mention, when you create the
We've been doing a lot of work on CephFS over the past few months. This
is an update on the current state of things as of Giant.
...
* Either the kernel client (kernel 3.17 or later) or userspace (ceph-fuse
or libcephfs) clients are in good working order.
Thanks for all the work and
hi christian,
we once were debugging some performance isssues, and IRQ balancing was
one of the issues we looked in, but no real benefit there for us.
all interrupts on one cpu is only an issue if the hardware itself is not
the bottleneck. we were running some default SAS HBA (Dell H200), and
but another issue is the OSD processes: do you pin those as well? and
how much data do they actually handle. to checksum, the OSD process
needs all data, so that can also cause a lot of NUMA traffic, esp if
they are not pinned.
That's why all my (production) storage nodes have only a single 6
for each OSD; can these be HT cores or actual physical cores?
stijn
Regards, Anand
-Original Message- From: ceph-users
[mailto:ceph-users-boun...@lists.ceph.com] On Behalf Of Stijn De
Weirdt Sent: Monday, September 22, 2014 2:36 PM To:
ceph-users@lists.ceph.com Subject: Re: [ceph-users
hi steven,
we ran into issues when trying to use a non-default user ceph user in
opennebula (don't remeber what the default was; but it's probably not
libvirt2 ), patches are in https://github.com/OpenNebula/one/pull/33,
devs sort-of confirmed they will be in 4.8.1. this way you can set
hi all,
there are a zillion OSD bug fixes. Things are looking pretty good for the
Giant release that is coming up in the next month.
any chance of having a compilable cephfs kernel module for el7 for the
next major release?
stijn
___
ceph-users
pool fails). but probably the read-cache has
to be the same as the write cache (eg when people want to modify a file).
stijn
On 07/08/2014 05:24 PM, Stijn De Weirdt wrote:
hi mark,
thanks for clarifying it a bit. we'll certainly have a look at the
caching tier setup.
stijn
On 07/08/2014 01
hi all,
one of the changes in the 0.82 release (accoridng to the notes) is:
mon: prevent EC pools from being used with cephfs
can someone clarify this a bit? cephfs with EC pools make no sense? now?
ever? or is it just not recommended (i'm also interested in the
technical reasons behind it)
hi mark,
thanks for clarifying it a bit. we'll certainly have a look at the
caching tier setup.
stijn
On 07/08/2014 01:53 PM, Mark Nelson wrote:
On 07/08/2014 04:28 AM, Stijn De Weirdt wrote:
hi all,
one of the changes in the 0.82 release (accoridng to the notes) is:
mon: prevent EC
that the kmod-ceph src rpms also provide
the ceph module.
stijn
On 06/27/2014 06:26 PM, Stijn De Weirdt wrote:
hi all,
does anyone know how to build the ceph.ko module for centos7
(3.10.0-123.el7 kernel) QA release? rebuilding the 0.81 ceph-kmod src
rpms gives modules for libceph and rbd, but the one
hi all,
does anyone know how to build the ceph.ko module for centos7
(3.10.0-123.el7 kernel) QA release? rebuilding the 0.81 ceph-kmod src
rpms gives modules for libceph and rbd, but the one for ceph fails with
error (same issue as for rhel7rc, see
hi kenneth,
yes there are if you create them. from the centos git sources wiki, use
the qemu-kvm repo, in the spec set rhev to 1 (and change the release)
and build it. update the installed rpms and done.
works out of the box (but maybe not so much from your side of our office ;)
stijn
On
this issue.
ok, we'll rebuild and try asap
stijn
Regards
Yan, Zheng
Thanks!
Regards
Yan, Zheng
Thanks!
Kenneth
- Message from Stijn De Weirdt stijn.dewei...@ugent.be -
Date: Fri, 04 Apr 2014 20:31:34 +0200
From: Stijn De Weirdt stijn.dewei...@ugent.be
Subject: Re: [ceph
hi yan,
(taking the list in CC)
On 04/04/2014 04:44 PM, Yan, Zheng wrote:
On Thu, Apr 3, 2014 at 2:52 PM, Stijn De Weirdt stijn.dewei...@ugent.be wrote:
hi,
latest pprof output attached.
this is no kernel client, this is ceph-fuse on EL6. starting the mds without
any ceph-fuse mounts works
hi gregory,
(i'm a colleague of kenneth)
1) How big and what shape the filesystem is. Do you have some
extremely large directory that the MDS keeps trying to load and then
dump?
anyway to extract this from the mds without having to start it? as it
was an rsync operation, i can try to locate
hi,
1) How big and what shape the filesystem is. Do you have some
extremely large directory that the MDS keeps trying to load and then
dump?
anyway to extract this from the mds without having to start it? as it
was an rsync operation, i can try to locate possible candidates on the
source
can we retest the clock skew condition? or get the value that the skew is?
ceph status gives
health HEALTH_WARN clock skew detected on mon.ceph003
in a polysh session (ie parallel ssh sort of thing)
ready (3) date +%s.%N
ceph002 : 1394713567.184218678
ceph003 : 1394713567.182722045
ceph001 :
we tried this with a Dell H200 (also LSI2008 based).
however, running some basic benchmarks, we saw no immediate difference
between IT and IR firmware.
so i'd like to know: what kind of performance improvement do you get,
and how did you measure it it?
thanks a lot
stijn
the howto for
we tried this with a Dell H200 (also LSI2008 based).
however, running some basic benchmarks, we saw no immediate difference
between IT and IR firmware.
so i'd like to know: what kind of performance improvement do you get,
and how did you measure it it?
IMO, flashing to IT firmware is mainly
39 matches
Mail list logo