[ceph-users] nautilus: dashboard configuration issue

Daniele Riccucci devster at posteo.net
Mon Mar 18 10:16:34 PDT 2019


Hello,
thank you for the fix, I'll be on the lookout for the next ceph/daemon 
container image.

Regards,
Daniele

On 18/03/19 12:59, Volker Theile wrote:
> Hello Daniele,
> 
> your problem is tracker by https://tracker.ceph.com/issues/38528 and
> fixed in the latest Ceph 14 builds. To workaround the problem simply
> disable SSL for your specific manager.
> 
> $ ceph config set mgr mgr/dashboard/<MGR_ID>/ssl false
> 
> See https://tracker.ceph.com/issues/38528#note-1.
> 
> 
> Regards
> Volker
> 
> 
> Am 17.03.19 um 03:11 schrieb Daniele Riccucci:
>> Hello,
>> I have a small cluster deployed with ceph-ansible on containers.
>> Recently, without realizing that ceph_docker_image_tag was set to
>> latest by default, the cluster got upgraded to nautilus and I was
>> unable to roll back.
>> Everything seems to be running smoothly except for the dashboard.
>>
>> $ ceph status
>>
>>    cluster:
>>      id:     d5c50302-0d8e-47cb-ab86-c15842372900
>>      health: HEALTH_ERR
>>              Module 'dashboard' has failed: IOError("Port 8443 not free
>> on '::'",)
>> [...]
>>
>> I already have a service running on port 8443 and I don't need SSL so
>> I ran:
>>
>>      ceph config set mgr mgr/dashboard/ssl false
>>
>> and
>>
>>      ceph config set mgr mgr/dashboard/server_port <myport>
>>
>> according to the docs, to change this behavior.
>> Running `ceph config dump` returns the following:
>>
>> WHO MASK LEVEL    OPTION                     VALUE    RO
>> mgr      advanced mgr/dashboard/server_port  8088     *
>> mgr      advanced mgr/dashboard/ssl          false    *
>>
>> By dumping the configuration I found that 2 keys were present:
>>
>> $ ceph config-key dump
>> {
>>      [...]
>>      "config/mgr/mgr/dashboard/server_port": "8088",
>>      "config/mgr/mgr/dashboard/ssl": "false",
>>      "config/mgr/mgr/dashboard/username": "devster",
>>      "mgr/dashboard/key": "<certificate>",
>>      "mgr/dashboard/crt": "<certificate>"
>> }
>>
>> which likely are the SSL certificates. I deleted them, disabled the
>> module and re-enabled it, however the following happened:
>>
>> $ ceph status
>>    cluster:
>>      id:     d5c50302-0d8e-47cb-ab86-c15842372900
>>      health: HEALTH_OK
>> [...]
>>
>> $ ceph mgr module ls | jq .enabled_modules
>> [
>>    "dashboard",
>>    "iostat",
>>    "prometheus",
>>    "restful"
>> ]
>>
>> $ ceph mgr services
>> {
>>    "prometheus": "http://localhost:9283/"
>> }
>>
>> Dashboard seems enabled but unavailable.
>> What am I missing?
>> Thank you.
>>
>> Daniele
>>
>> P.S. what is the difference between `ceph config` and `ceph config-key`?
>> _______________________________________________
>> ceph-users mailing list
>> ceph-users at lists.ceph.com
>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
>>
>>
>> _______________________________________________
>> ceph-users mailing list
>> ceph-users at lists.ceph.com
>> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com


More information about the ceph-users mailing list