[one-users] Ceph datastore monitoring in OpenNebula
Alvaro Simon Garcia
Alvaro.SimonGarcia at UGent.be
Mon Aug 11 01:22:10 PDT 2014
Hi Jaime, Javier
Do you think that it would be possible to include these ceph patches
into next 4.8 release? At this moment we are applying theses patches by
hand in our configuration to use different Ceph pools.
https://github.com/OpenNebula/one/pull/27
https://github.com/OpenNebula/one/pull/28
https://github.com/OpenNebula/one/pull/29
https://github.com/OpenNebula/one/pull/30
https://github.com/OpenNebula/one/pull/31
Cheers
Alvaro
On 17/07/14 14:11, Alvaro Simon Garcia wrote:
> Hi Jaime
>
> Sorry for the late reply, I didn't see your mail before because I was
> included in cc. With this patch you don't need to include a new user
> into your ceph keyring. If you want to use a ceph datastore you only
> need to include the user and pool into the datastore template that's
> all. In our case we use livbirt datastore and we have created two ceph
> datastore for testing porposes (each one use a different pool as well):
>
>
> $ onedatastore show ceph
> DATASTORE 103 INFORMATION
> ID : 103
> NAME : ceph
> USER : oneadmin
> GROUP : oneadmin
> CLUSTER : -
> TYPE : IMAGE
> DS_MAD : ceph
> TM_MAD : ceph
> BASE PATH : /var/lib/one//datastores/103
> DISK_TYPE : RBD
>
> DATASTORE CAPACITY
> TOTAL: : 87.6T
> FREE: : 59.2T
> USED: : 28.4T
> LIMIT: : -
>
> PERMISSIONS
> OWNER : um-
> GROUP : u--
> OTHER : ---
>
> DATASTORE TEMPLATE
> BASE_PATH="/var/lib/one//datastores/"
> BRIDGE_LIST="hyp004.cubone.os"
> CEPH_HOST="ceph001.cubone.os ceph002.cubone.os ceph003.cubone.os"
> CEPH_SECRET="xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
> CEPH_USER="libvirt"
> CLONE_TARGET="SELF"
> DATASTORE_CAPACITY_CHECK="yes"
> DISK_TYPE="RBD"
> DS_MAD="ceph"
> LN_TARGET="NONE"
> POOL_NAME="one"
> TM_MAD="ceph"
>
> IMAGES
> 7
> 8
> 31
> 32
> 34
> 35
>
>
> and the second one:
>
> $ onedatastore show "ceph two"
> DATASTORE 104 INFORMATION
> ID : 104
> NAME : ceph two
> USER : oneadmin
> GROUP : oneadmin
> CLUSTER : -
> TYPE : IMAGE
> DS_MAD : ceph
> TM_MAD : ceph
> BASE PATH : /var/lib/one//datastores/104
> DISK_TYPE : RBD
>
> DATASTORE CAPACITY
> TOTAL: : 87.6T
> FREE: : 59.2T
> USED: : 28.4T
> LIMIT: : -
>
> PERMISSIONS
> OWNER : um-
> GROUP : u--
> OTHER : ---
>
> DATASTORE TEMPLATE
> BASE_PATH="/var/lib/one//datastores/"
> BRIDGE_LIST="hyp004.cubone.os"
> CEPH_HOST="ceph001.cubone.os ceph002.cubone.os ceph003.cubone.os"
> CEPH_SECRET="xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
> CEPH_USER="libvirt"
> CLONE_TARGET="SELF"
> DATASTORE_CAPACITY_CHECK="yes"
> DISK_TYPE="RBD"
> DS_MAD="ceph"
> LN_TARGET="NONE"
> POOL_NAME="two"
> TM_MAD="ceph"
>
> As you can see we are using different pools in each one (so we don't
> need to include the pool name into ceph.conf either, and we are able
> to use several ceph clusters) and this change simplifies the ONE conf
> and ceph cluster admin as well.
>
> Cheers
> Alvaro
>
> On 03/07/14 17:08, Jaime Melis wrote:
>> Hi Alvaro,
>>
>> thanks a lot for the contribution. I haven't tested it yet, but the
>> code seems to be perfect.
>>
>> My main issue here is that I'm not entirely sure if we need this.
>> Doesn't it make sense for the oneadmin user to be in the keyring and
>> be able to run rados? You mentioned this, but I'm having some
>> trouble understanding what do you mean:
>>
>> > This feature could be useful as well if you want to monitor several
>> datastores that are using different ceph pools and users ids.
>> > You only have to include the id and pool info into the ONE
>> datastore template and the monitoring script will use one or another
>> depending on the DS conf.
>>
>> With the current system you can monitor multiple ceph pools as long
>> as the oneadmin user has right, isn't that so?
>>
>> Joel, would you like to weigh in? would you find this useful?
>>
>> cheers,
>> Jaime
>>
>>
>> On Fri, Jun 20, 2014 at 3:23 PM, Javier Fontan
>> <jfontan at opennebula.org <mailto:jfontan at opennebula.org>> wrote:
>>
>> Great, thanks! Our ceph dev is traveling right now, he will
>> review the
>> patches when he arrives.
>>
>> Cheers
>>
>> On Thu, Jun 19, 2014 at 3:21 PM, Alvaro Simon Garcia
>> <Alvaro.SimonGarcia at ugent.be
>> <mailto:Alvaro.SimonGarcia at ugent.be>> wrote:
>> > Hi Javier
>> >
>> > We have modified the ceph monitor to take into account the
>> datastore ceph
>> > pool and also the ceph user. This is a generic solution that
>> could be useful
>> > for other datacenters as well, we have created a pull request
>> in github if
>> > you are agree about this change and you want to include it in
>> the next
>> > release.
>> >
>> > https://github.com/OpenNebula/one/pull/27
>> >
>> > We only have modified these lines in
>> > /var/lib/one/remotes/datastore/ceph/monitor:
>> >
>> >> --- monitor.orig.190614 2014-06-19 14:35:24.022755989 +0200
>> >> +++ monitor 2014-06-19 14:49:34.043187892 +0200
>> >> @@ -46,10 +46,12 @@
>> >> while IFS= read -r -d '' element; do
>> >> XPATH_ELEMENTS[i++]="$element"
>> >> done < <($XPATH
>> /DS_DRIVER_ACTION_DATA/DATASTORE/TEMPLATE/BRIDGE_LIST \
>> >> - /DS_DRIVER_ACTION_DATA/DATASTORE/TEMPLATE/POOL_NAME)
>> >> + /DS_DRIVER_ACTION_DATA/DATASTORE/TEMPLATE/POOL_NAME \
>> >> + /DS_DRIVER_ACTION_DATA/DATASTORE/TEMPLATE/CEPH_USER)
>> >>
>> >> BRIDGE_LIST="${XPATH_ELEMENTS[j++]}"
>> >> POOL_NAME="${XPATH_ELEMENTS[j++]:-$POOL_NAME}"
>> >> +CEPH_USER="${XPATH_ELEMENTS[j++]}"
>> >>
>> >> HOST=`get_destination_host`
>> >>
>> >> @@ -61,7 +63,7 @@
>> >> # ------------ Compute datastore usage -------------
>> >>
>> >> MONITOR_SCRIPT=$(cat <<EOF
>> >> -$RADOS df | $AWK '{
>> >> +$RADOS df -p ${POOL_NAME} --id ${CEPH_USER}| $AWK '{
>> >> if (\$1 == "total") {
>> >>
>> >> space = int(\$3/1024)
>> >
>> >
>> > CEPH_USER and POOL_NAME should be mandatory to create the ceph
>> datastore.
>> >
>> > Cheers
>> > Alvaro
>> >
>> >
>> >
>> >> Hola Javi
>> >>
>> >> Thanks a lot for your feedback. Yes we will modify the current
>> >> monitoring scripts to take into account this. This feature
>> could be
>> >> useful as well if you want to monitor several datastores that
>> are using
>> >> different ceph pools and users ids. You only have to include
>> the id and
>> >> pool info into the ONE datastore template and the monitoring
>> script will
>> >> use one or another depending on the DS conf.
>> >>
>> >>
>> >> Cheers and thanks!
>> >> Alvaro
>> >> On 2014-06-17 14:55, Javier Fontan wrote:
>> >>>
>> >>> CEPH_USER is used when generating the libvirt/kvm deployment
>> file but
>> >>> not for DS monitoring:
>> >>>
>> >>> * Deployment file generation:
>> >>>
>> >>>
>> https://github.com/OpenNebula/one/blob/one-4.6/src/vmm/LibVirtDriverKVM.cc#L461
>> >>> * Monitoring:
>> >>>
>> https://github.com/OpenNebula/one/blob/one-4.6/src/datastore_mad/remotes/ceph/monitor#L64
>> >>>
>> >>> Ceph is not may area of expertise but you may need to add those
>> >>> parameters to the monitor script a maybe to other scripts
>> that use the
>> >>> "rados" command. It may also be possible to modify the RADOS
>> command
>> >>> to have those parameters instead of modifying all the scripts:
>> >>>
>> >>>
>> >>>
>> https://github.com/OpenNebula/one/blob/one-4.6/src/mad/sh/scripts_common.sh#L40
>> >>>
>> >>> As I said I don't know much about Ceph and it may be those
>> credentials
>> >>> could be set in a config file or so.
>> >>>
>> >>> On Tue, Jun 17, 2014 at 11:19 AM, Alvaro Simon Garcia
>> >>> <Alvaro.SimonGarcia at ugent.be
>> <mailto:Alvaro.SimonGarcia at ugent.be>> wrote:
>> >>>>
>> >>>> Hi
>> >>>>
>> >>>> We have included the admin keyring instead of libvirt user
>> and it
>> >>>> works...
>> >>>> that means that we can run rbd or qemu-img wihtout the
>> libvirt id, but
>> >>>> is
>> >>>> not the best solution. We have included the user into
>> datastore conf:
>> >>>>
>> >>>> CEPH_USER="libvirt"
>> >>>>
>> >>>> but it seems that is not used by opennebula at the end
>> >>>>
>> >>>> Cheers
>> >>>> Alvaro
>> >>>>
>> >>>>
>> >>>> On 2014-06-17 10:09, Alvaro Simon Garcia wrote:
>> >>>>>
>> >>>>> Hi all
>> >>>>>
>> >>>>>
>> >>>>> We have included our ONE nodes into Ceph cluster, cephx
>> auth is working
>> >>>>> but OpenNebula is not able to detect the free space:
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>>> $ onedatastore show 103
>> >>>>>> DATASTORE 103 INFORMATION
>> >>>>>> ID : 103
>> >>>>>> NAME : ceph
>> >>>>>> USER : oneadmin
>> >>>>>> GROUP : oneadmin
>> >>>>>> CLUSTER : -
>> >>>>>> TYPE : IMAGE
>> >>>>>> DS_MAD : ceph
>> >>>>>> TM_MAD : ceph
>> >>>>>> BASE PATH : /var/lib/one//datastores/103
>> >>>>>> DISK_TYPE : RBD
>> >>>>>>
>> >>>>>> DATASTORE CAPACITY
>> >>>>>> TOTAL: : 0M
>> >>>>>> FREE: : 0M
>> >>>>>> USED: : 0M
>> >>>>>> LIMIT: : -
>> >>>>>>
>> >>>>>> PERMISSIONS
>> >>>>>> OWNER : um-
>> >>>>>> GROUP : u--
>> >>>>>> OTHER : ---
>> >>>>>
>> >>>>>
>> >>>>>> $ onedatastore list
>> >>>>>> ID NAME SIZE AVAIL CLUSTER IMAGES TYPE DS
>> >>>>>> TM
>> >>>>>> 0 system 114.8G 85% - 0 sys -
>> >>>>>> shared
>> >>>>>> 1 default 114.9G 84% - 2 img fs
>> >>>>>> ssh
>> >>>>>> 2 files 114.9G 84% - 0 fil fs
>> >>>>>> ssh
>> >>>>>> 103 ceph 0M - - 0 img ceph
>> >>>>>> ceph
>> >>>>>
>> >>>>>
>> >>>>> but if we run rados as oneadmin user:
>> >>>>>
>> >>>>>> $ rados df -p one --id libvirt
>> >>>>>> pool name category KB objects clones
>> >>>>>> degraded unfound rd rd KB wr wr
>> >>>>>> KB
>> >>>>>> one - 0 0 0
>> >>>>>> 0 0 0 0 0 0
>> >>>>>> total used 1581852 37
>> >>>>>> total avail 140846865180
>> >>>>>> total space 140848447032
>> >>>>>
>> >>>>>
>> >>>>> It's working correctly (we are using one pool and libvirt
>> ceph id)
>> >>>>>
>> >>>>> the oned.log only shows this info:
>> >>>>> Tue Jun 17 10:06:37 2014 [InM][D]: Monitoring datastore
>> default (1)
>> >>>>> Tue Jun 17 10:06:37 2014 [InM][D]: Monitoring datastore
>> files (2)
>> >>>>> Tue Jun 17 10:06:37 2014 [InM][D]: Monitoring datastore
>> ceph (103)
>> >>>>> Tue Jun 17 10:06:37 2014 [ImM][D]: Datastore default (1)
>> successfully
>> >>>>> monitored.
>> >>>>> Tue Jun 17 10:06:37 2014 [ImM][D]: Datastore files (2)
>> successfully
>> >>>>> monitored.
>> >>>>> Tue Jun 17 10:06:37 2014 [ImM][D]: Datastore ceph (103)
>> successfully
>> >>>>> monitored.
>> >>>>>
>> >>>>> Any clue about how to debug this issue?
>> >>>>>
>> >>>>> Thanks in advance!
>> >>>>> Alvaro
>> >>>>>
>> >>>>>
>> >>>>>
>> >>>>> _______________________________________________
>> >>>>> Users mailing list
>> >>>>> Users at lists.opennebula.org <mailto:Users at lists.opennebula.org>
>> >>>>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>> >>>>
>> >>>> _______________________________________________
>> >>>> Users mailing list
>> >>>> Users at lists.opennebula.org <mailto:Users at lists.opennebula.org>
>> >>>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>> >>>
>> >>>
>> >> _______________________________________________
>> >> Users mailing list
>> >> Users at lists.opennebula.org <mailto:Users at lists.opennebula.org>
>> >> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>> >
>> >
>>
>>
>>
>> --
>> Javier Fontán Muiños
>> Developer
>> OpenNebula - Flexible Enterprise Cloud Made Simple
>> www.OpenNebula.org <http://www.OpenNebula.org> | @OpenNebula |
>> github.com/jfontan <http://github.com/jfontan>
>> _______________________________________________
>> Users mailing list
>> Users at lists.opennebula.org <mailto:Users at lists.opennebula.org>
>> http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>>
>>
>>
>>
>> --
>> Jaime Melis
>> Project Engineer
>> OpenNebula - Flexible Enterprise Cloud Made Simple
>> www.OpenNebula.org <http://www.OpenNebula.org> |
>> jmelis at opennebula.org <mailto:jmelis at opennebula.org>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20140811/e0c143a9/attachment-0001.htm>
More information about the Users
mailing list