There is ‘cinder-manage db purge’ to delete entries which are marked as deleted.

I never tried it (and suggest to start with a copy of the db before touching 
the prod
ones). We usually do not purge databases unless we hit an issue.

Cheers,
 Arne


> On 02 Sep 2016, at 10:28, William Josefsson <[email protected]> 
> wrote:
> 
> Thanks everyone for your replies! I did a safe select first to make
> sure there was only one match. than I updated deleted=1 for that
> service which seem to work. Now 'cinder service-list' shows the right
> output.
> 
> I notice in DB 'volumes', there are plenty of old volume entries, long
> ago deleted, and they have 'deleted=1'. The host value, is the old
> host name that no longer exist.
> 
> Is there any cleanup of volumes entries with deleted=1, or is it
> normal these old entries lay around? thx will
> 
> 
> 
> On Fri, Sep 2, 2016 at 1:24 AM, Kris G. Lindgren <[email protected]> 
> wrote:
>> Just be careful with LIMIT x on your servers if you have replicated mysql 
>> databases.  At least under older versions of mysql this can lead to broken 
>> replication as the results of the query performed on the master and on the 
>> slave are not guaranteed to be the same.
>> 
>> https://dev.mysql.com/doc/refman/5.7/en/replication-features-limit.html
>> 
>> ___________________________________________________________________
>> Kris Lindgren
>> Senior Linux Systems Engineer
>> GoDaddy
>> 
>> On 9/1/16, 9:51 AM, "Nick Jones" <[email protected]> wrote:
>> 
>> 
>>    On 1 Sep 2016, at 15:36, Jonathan D. Proulx wrote:
>> 
>>> On Thu, Sep 01, 2016 at 04:25:25PM +0300, Vladimir Prokofev wrote:
>>> :I've used direct database update to achive this in Mitaka:
>>> :use cinder;
>>> :update services set deleted = '1' where <your service parameters>;
>>> 
>>> 
>>> I belive the official way is:
>>> 
>>> cinder-manage service remove <binary> <host>
>>> 
>>> Which probably more or less does the same thing...
>> 
>>    Yep.  Both options basically require direct interaction with the
>>    database as opposed to via a Cinder API call, but at least with
>>    cinder-manage the scope for making a mistake is far more limited than
>>    missing some qualifying clause off an UPDATE statement (limit 1 is your
>>    friend!) ;)
>> 
>>    —
>> 
>>    -Nick
>> 
>>    --
>>    DataCentred Limited registered in England and Wales no. 05611763
>> 
>>    _______________________________________________
>>    OpenStack-operators mailing list
>>    [email protected]
>>    http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
>> 
>> 
>> _______________________________________________
>> OpenStack-operators mailing list
>> [email protected]
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
> 
> _______________________________________________
> OpenStack-operators mailing list
> [email protected]
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators

--
Arne Wiebalck
CERN IT

_______________________________________________
OpenStack-operators mailing list
[email protected]
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators

Reply via email to