[PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Jean-Luc Oms]

Bonjour,

Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
VE seems to introduce a python 2/3 version problem, dashboard healt
stops working.

root@ceph1:/usr/bin# ceph health
HEALTH_ERR Module 'dashboard' has failed: ('invalid syntax',
('/usr/share/ceph/mgr/dashboard/controllers/orchestrator.py', 34, 11,
'    result: dict = {}\n'))

This syntax was introduced in 3.6, and using strace it seems python 2.7
is used.

Any option to resolve this ? everything was ok in 6.2-15.

Thanks


-- 
Jean-Luc Oms
/STI-ReseauX <https://rx.lirmm.fr>- LIRMM - CNRS/UM/
+33 4 67 41 85 93 <tel:+33-467-41-85-93> / +33 6 32 01 04 17
<tel:+33-632-01-04-17>

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Lindsay Mathieson]

On 27/11/2020 9:47 pm, Jean-Luc Oms wrote:
> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
> VE seems to introduce a python 2/3 version problem, dashboard healt
> stops working.

Was just about to report that :)


Same here.

-- 
Lindsay

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[alexandre derumier]

 >>1 pools have too many placement groups Pool rbd has 128 placement 
groups, should have 32
 >>
 >>I don't find any way to reduce placement groups tu 32
 >>
 >>Any help welcome.

you can't reduce PG on nautilus,  only since octopus.  (and ceph is able 
to do it automataaly with new pg autoscaler)

I think it's a warning introduced in last nautilus update.

If I remember, they are an option to disable this warning, (but I don't 
remember it)


On 27/11/2020 17:29, Jean-Daniel TISSOT wrote:
> Hi,
>
> I have another problem
>
> root@dmz-pve1:~ # ceph health HEALTH_WARN 1 pools have too many 
> placement groups root@dmz-pve1:~ # pveceph pool ls 
> ┌───────────────────────┬──────┬──────────┬────────┬───────────────────┬─────────────────┬──────────────────────┬──────────────┐ 
> │ Name                  │ Size │ Min Size │ PG Num │ PG Autoscale Mode 
> │ Crush Rule Name │               %-Used │         Used │ 
> ╞═══════════════════════╪══════╪══════════╪════════╪═══════════════════╪═════════════════╪══════════════════════╪══════════════╡ 
> │ device_health_metrics │    3 │        2 │      1 │ on                
> │ replicated_rule │ 4.19273845864154e-07 │ 4534827 │ 
> ├───────────────────────┼──────┼──────────┼────────┼───────────────────┼─────────────────┼──────────────────────┼──────────────┤ 
> │ rbd                   │    3 │        2 │    128 │ warn              
> │ replicated_rule │   0.0116069903597236 │ 127014329075 │ 
> └───────────────────────┴──────┴──────────┴────────┴───────────────────┴─────────────────┴──────────────────────┴──────────────┘ 
>
>
> In the GUI :
>
> 1 pools have too many placement groups Pool rbd has 128 placement 
> groups, should have 32
>
> I don't find any way to reduce placement groups tu 32
>
> Any help welcome.
>
> Best regards,
>
> Le 27/11/2020 à 14:07, Lindsay Mathieson a écrit :
>> On 27/11/2020 9:47 pm, Jean-Luc Oms wrote:
>>> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
>>> VE seems to introduce a python 2/3 version problem, dashboard healt
>>> stops working.
>>
>> Was just about to report that :)
>>
>>
>> Same here.
>>

[PVE-User] ProxmoxVE6.3/CEPH Octopus 15.2.6 was Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Jean-Daniel TISSOT]

Sorry to steal the thread.

In fact before upgrading to Octopus, I don't remember any warning.

I upgrade Proxmox an follow the wiki to upgrade CEPH. All things seam 
working (I don't test HA by migrate a VM work perfectly.

I have just the warning on pool rdb (1 pools have too many placement 
groups Pool rbd has 128 placement groups, should have 32)


Le 27/11/2020 à 17:59, alexandre derumier a écrit :
> >>1 pools have too many placement groups Pool rbd has 128 placement 
> groups, should have 32
> >>
> >>I don't find any way to reduce placement groups tu 32
> >>
> >>Any help welcome.
>
> you can't reduce PG on nautilus,  only since octopus.  (and ceph is 
> able to do it automataaly with new pg autoscaler)
>
> I think it's a warning introduced in last nautilus update.
>
> If I remember, they are an option to disable this warning, (but I 
> don't remember it)
>
>
> On 27/11/2020 17:29, Jean-Daniel TISSOT wrote:
>> Hi,
>>
>> I have another problem
>>
>> root@dmz-pve1:~ # ceph health HEALTH_WARN 1 pools have too many 
>> placement groups root@dmz-pve1:~ # pveceph pool ls 
>> ┌───────────────────────┬──────┬──────────┬────────┬───────────────────┬─────────────────┬──────────────────────┬──────────────┐ 
>> │ Name                  │ Size │ Min Size │ PG Num │ PG Autoscale 
>> Mode │ Crush Rule Name │               %-Used │         Used │ 
>> ╞═══════════════════════╪══════╪══════════╪════════╪═══════════════════╪═════════════════╪══════════════════════╪══════════════╡ 
>> │ device_health_metrics │    3 │        2 │      1 │ 
>> on                │ replicated_rule │ 4.19273845864154e-07 │ 4534827 
>> │ 
>> ├───────────────────────┼──────┼──────────┼────────┼───────────────────┼─────────────────┼──────────────────────┼──────────────┤ 
>> │ rbd                   │    3 │        2 │    128 │ 
>> warn              │ replicated_rule │   0.0116069903597236 │ 
>> 127014329075 │ 
>> └───────────────────────┴──────┴──────────┴────────┴───────────────────┴─────────────────┴──────────────────────┴──────────────┘ 
>>
>>
>> In the GUI :
>>
>> 1 pools have too many placement groups Pool rbd has 128 placement 
>> groups, should have 32
>>
>> I don't find any way to reduce placement groups tu 32
>>
>> Any help welcome.
>>
>> Best regards,
>>
>> Le 27/11/2020 à 14:07, Lindsay Mathieson a écrit :
>>> On 27/11/2020 9:47 pm, Jean-Luc Oms wrote:
>>>> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
>>>> VE seems to introduce a python 2/3 version problem, dashboard healt
>>>> stops working.
>>>
>>> Was just about to report that :)
>>>
>>>
>>> Same here.
>>>
>
>
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

-- 
Bien cordialement,
Jean-Daniel Tissot - IE CNRS     http://chrono-environnement.univ-fcomte.fr
UMR 6249 - Laboratoire Chrono-environnement                    UMR CNRS-UFC
Université de Franche-Comté, 16 route de Gray, 25030 Besançon Cedex, FRANCE
Jean-Daniel.Tissot@univ-fcomte.fr                      tel:+33 3 81 666 440

Alabama, Mississippi, Minnesota, South Carolina, Oregon... not so sweet home
Black Panther Party, renaît de tes cendres et reviens les aider
https://www.youtube.com/watch?v=ZvilFSMVHTs

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Lindsay Mathieson]

On 28/11/2020 2:59 am, alexandre derumier wrote:
> you can't reduce PG on nautilus,  only since octopus.  (and ceph is 
> able to do it automataaly with new pg autoscaler)


Actually you can, has been the case since Nautilus.


https://ceph.io/rados/new-in-nautilus-pg-merging-and-autotuning/


-- 
Lindsay

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Marco M. Gabriel]

You can enable the pg-autoscaler. It does the work for you and
optimizes the number of placement groups on a given pool.

The pg-autoscaler was introduced with nautilus and we ran it for a
while without any problems.

Here is an explanation of how to enable and how to use the autoscaler:
https://ceph.io/rados/new-in-nautilus-pg-merging-and-autotuning/

Best regards,
Marco


Am Fr., 27. Nov. 2020 um 17:37 Uhr schrieb Jean-Daniel TISSOT
<Jean-Daniel.Tissot@univ-fcomte.fr>:
>
> Hi,
>
> I have another problem
>
> root@dmz-pve1:~ # ceph health HEALTH_WARN 1 pools have too many
> placement groups root@dmz-pve1:~ # pveceph pool ls
> ┌───────────────────────┬──────┬──────────┬────────┬───────────────────┬─────────────────┬──────────────────────┬──────────────┐
> │ Name                  │ Size │ Min Size │ PG Num │ PG Autoscale Mode │
> Crush Rule Name │               %-Used │         Used │
> ╞═══════════════════════╪══════╪══════════╪════════╪═══════════════════╪═════════════════╪══════════════════════╪══════════════╡
> │ device_health_metrics │    3 │        2 │      1 │ on                │
> replicated_rule │ 4.19273845864154e-07 │      4534827 │
> ├───────────────────────┼──────┼──────────┼────────┼───────────────────┼─────────────────┼──────────────────────┼──────────────┤
> │ rbd                   │    3 │        2 │    128 │ warn              │
> replicated_rule │   0.0116069903597236 │ 127014329075 │
> └───────────────────────┴──────┴──────────┴────────┴───────────────────┴─────────────────┴──────────────────────┴──────────────┘
>
>
> In the GUI :
>
> 1 pools have too many placement groups Pool rbd has 128 placement
> groups, should have 32
>
> I don't find any way to reduce placement groups tu 32
>
> Any help welcome.
>
> Best regards,
>
> Le 27/11/2020 à 14:07, Lindsay Mathieson a écrit :
> > On 27/11/2020 9:47 pm, Jean-Luc Oms wrote:
> >> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
> >> VE seems to introduce a python 2/3 version problem, dashboard healt
> >> stops working.
> >
> > Was just about to report that :)
> >
> >
> > Same here.
> >
> --
> Bien cordialement,
> Jean-Daniel Tissot - IE CNRS     http://chrono-environnement.univ-fcomte.fr
> UMR 6249 - Laboratoire Chrono-environnement                    UMR CNRS-UFC
> Université de Franche-Comté, 16 route de Gray, 25030 Besançon Cedex, FRANCE
> Jean-Daniel.Tissot@univ-fcomte.fr                      tel:+33 3 81 666 440
>
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[alexandre derumier]

On 27/11/2020 18:06, Marco M. Gabriel wrote:
> You can enable the pg-autoscaler. It does the work for you and
> optimizes the number of placement groups on a given pool.

Oh, yes indeed, I thinked it was introduced in Octopus,

but it's indeed already avaible in Nautilus :)

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Jean-Daniel TISSOT]

Many thanks.

Work like a charm.

No more warning.

Again many thanks Marco

Best regards,
Jean-Daniel

Le 27/11/2020 à 18:06, Marco M. Gabriel a écrit :
> You can enable the pg-autoscaler. It does the work for you and
> optimizes the number of placement groups on a given pool.
>
> The pg-autoscaler was introduced with nautilus and we ran it for a
> while without any problems.
>
> Here is an explanation of how to enable and how to use the autoscaler:
> https://ceph.io/rados/new-in-nautilus-pg-merging-and-autotuning/
>
> Best regards,
> Marco
>
>
> Am Fr., 27. Nov. 2020 um 17:37 Uhr schrieb Jean-Daniel TISSOT
> <Jean-Daniel.Tissot@univ-fcomte.fr>:
>> Hi,
>>
>> I have another problem
>>
>> root@dmz-pve1:~ # ceph health HEALTH_WARN 1 pools have too many
>> placement groups root@dmz-pve1:~ # pveceph pool ls
>> ┌───────────────────────┬──────┬──────────┬────────┬───────────────────┬─────────────────┬──────────────────────┬──────────────┐
>> │ Name                  │ Size │ Min Size │ PG Num │ PG Autoscale Mode │
>> Crush Rule Name │               %-Used │         Used │
>> ╞═══════════════════════╪══════╪══════════╪════════╪═══════════════════╪═════════════════╪══════════════════════╪══════════════╡
>> │ device_health_metrics │    3 │        2 │      1 │ on                │
>> replicated_rule │ 4.19273845864154e-07 │      4534827 │
>> ├───────────────────────┼──────┼──────────┼────────┼───────────────────┼─────────────────┼──────────────────────┼──────────────┤
>> │ rbd                   │    3 │        2 │    128 │ warn              │
>> replicated_rule │   0.0116069903597236 │ 127014329075 │
>> └───────────────────────┴──────┴──────────┴────────┴───────────────────┴─────────────────┴──────────────────────┴──────────────┘
>>
>>
>> In the GUI :
>>
>> 1 pools have too many placement groups Pool rbd has 128 placement
>> groups, should have 32
>>
>> I don't find any way to reduce placement groups tu 32
>>
>> Any help welcome.
>>
>> Best regards,
>>
>> Le 27/11/2020 à 14:07, Lindsay Mathieson a écrit :
>>> On 27/11/2020 9:47 pm, Jean-Luc Oms wrote:
>>>> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
>>>> VE seems to introduce a python 2/3 version problem, dashboard healt
>>>> stops working.
>>> Was just about to report that :)
>>>
>>>
>>> Same here.
>>>
>> --
>> Bien cordialement,
>> Jean-Daniel Tissot - IE CNRS     http://chrono-environnement.univ-fcomte.fr
>> UMR 6249 - Laboratoire Chrono-environnement                    UMR CNRS-UFC
>> Université de Franche-Comté, 16 route de Gray, 25030 Besançon Cedex, FRANCE
>> Jean-Daniel.Tissot@univ-fcomte.fr                      tel:+33 3 81 666 440
>>
>> _______________________________________________
>> pve-user mailing list
>> pve-user@lists.proxmox.com
>> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

-- 
Bien cordialement,
Jean-Daniel Tissot - IE CNRS     http://chrono-environnement.univ-fcomte.fr
UMR 6249 - Laboratoire Chrono-environnement                    UMR CNRS-UFC
Université de Franche-Comté, 16 route de Gray, 25030 Besançon Cedex, FRANCE
Jean-Daniel.Tissot@univ-fcomte.fr                      tel:+33 3 81 666 440

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Jean-Luc Oms]

Next step ....

I've a small 'preprod' cluster for testing ... but without ceph. If I
install ceph on oe node of this cluster this package is not installed:
ceph-mgr-dashboard

If I remove this package from my prod cluster, tested on the node
running active manager, no dependencies and after manager restart,
health is Ok ...

Now i have installed :

root@ceph1:~# dpkg -l | grep ceph
ii  ceph                                 14.2.15-pve2                
amd64        distributed storage and file system
ii  ceph-base                            14.2.15-pve2                
amd64        common ceph daemon libraries and management tools
ii  ceph-common                          14.2.15-pve2                
amd64        common utilities to mount and interact with a ceph storage
cluster
ii  ceph-fuse                            14.2.15-pve2                
amd64        FUSE-based client for the Ceph distributed file system
ii  ceph-mds                             14.2.15-pve2                
amd64        metadata server for the ceph distributed file system
ii  ceph-mgr                             14.2.15-pve2                
amd64        manager for the ceph distributed storage system
ii  ceph-mon                             14.2.15-pve2                
amd64        monitor server for the ceph storage system
ii  ceph-osd                             14.2.15-pve2                
amd64        OSD server for the ceph storage system
ii  libcephfs2                           14.2.15-pve2                
amd64        Ceph distributed file system client library
ii  python-ceph-argparse                 14.2.15-pve2                
all          Python 2 utility libraries for Ceph CLI
ii  python-cephfs                        14.2.15-pve2                
amd64        Python 2 libraries for the Ceph libcephfs library

Is this ok ?

Is ceph-mgr-dashboard  needed ?

Thanks


Le 27/11/2020 à 12:47, Jean-Luc Oms a écrit :
> Bonjour,
>
> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
> VE seems to introduce a python 2/3 version problem, dashboard healt
> stops working.
>
> root@ceph1:/usr/bin# ceph health
> HEALTH_ERR Module 'dashboard' has failed: ('invalid syntax',
> ('/usr/share/ceph/mgr/dashboard/controllers/orchestrator.py', 34, 11,
> '    result: dict = {}\n'))
>
> This syntax was introduced in 3.6, and using strace it seems python 2.7
> is used.
>
> Any option to resolve this ? everything was ok in 6.2-15.
>
> Thanks
>
>
-- 
Jean-Luc Oms
/STI-ReseauX <https://rx.lirmm.fr>- LIRMM - CNRS/UM/
+33 4 67 41 85 93 <tel:+33-467-41-85-93> / +33 6 32 01 04 17
<tel:+33-632-01-04-17>

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Marco M. Gabriel]

Same problem here after upgrading from 6.2.15 to 6.3 on a test cluster.

But the problem disappeared suddenly when I upgraded Ceph from
Nautilus to Octopus as well. Not sure if this is the reason why it
disappeared and I wouldn't recommend doing this on a production
cluster while ceph is in HEALTH_ERR.

It would be fine if anyone could test and confirm that an upgrade to
Ceph Octopus resolves the issue.

Kind regards,
Marco

Am Fr., 27. Nov. 2020 um 12:54 Uhr schrieb Jean-Luc Oms <jean-luc.oms@lirmm.fr>:
>
> Bonjour,
>
> Upgrading to last Proxmox VE / Ceph nautilus from the last 6.2 proxmox
> VE seems to introduce a python 2/3 version problem, dashboard healt
> stops working.
>
> root@ceph1:/usr/bin# ceph health
> HEALTH_ERR Module 'dashboard' has failed: ('invalid syntax',
> ('/usr/share/ceph/mgr/dashboard/controllers/orchestrator.py', 34, 11,
> '    result: dict = {}\n'))
>
> This syntax was introduced in 3.6, and using strace it seems python 2.7
> is used.
>
> Any option to resolve this ? everything was ok in 6.2-15.
>
> Thanks
>
>
> --
> Jean-Luc Oms
> /STI-ReseauX <https://rx.lirmm.fr>- LIRMM - CNRS/UM/
> +33 4 67 41 85 93 <tel:+33-467-41-85-93> / +33 6 32 01 04 17
> <tel:+33-632-01-04-17>
> _______________________________________________
> pve-user mailing list
> pve-user@lists.proxmox.com
> https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Re: [PVE-User] Python problem with upgrade to proxmoxVE6.3/ CEPH Nautilus 14.2.15

[Lindsay Mathieson]

On 28/11/2020 2:45 am, Marco M. Gabriel wrote:
> It would be fine if anyone could test and confirm that an upgrade to
> Ceph Octopus resolves the issue.

Regards the dashboard not working on Proxmox 6.3 - upgrading to ceph 
Octopus fixed it for me. Also the dashboard looks a lot more swish :)

-- 
Lindsay