[ovirt-devel] [DB] Diffrent UUIDs for inserted data per installation
emesika at redhat.com
Thu Jan 29 13:31:45 UTC 2015
----- Original Message -----
> From: "Lior Vernia" <lvernia at redhat.com>
> To: "Moti Asayag" <masayag at redhat.com>
> Cc: "Martin Mucha" <mmucha at redhat.com>, "engine-devel at ovirt.org" <devel at ovirt.org>, "Eli Mesika" <emesika at redhat.com>
> Sent: Thursday, January 29, 2015 2:46:40 PM
> Subject: Re: [ovirt-devel] [DB] Diffrent UUIDs for inserted data per installation
> Apologies for the late response...
> On 25/01/15 13:44, Moti Asayag wrote:
> > ----- Original Message -----
> >> From: "Eli Mesika" <emesika at redhat.com>
> >> To: "Sahina Bose" <sabose at redhat.com>, "Allon Mureinik"
> >> <amureini at redhat.com>, "Arik Hadas" <ahadas at redhat.com>,
> >> "Mike Kolesnik" <mkolesni at redhat.com>, "Gilad Chaplik"
> >> <gchaplik at redhat.com>
> >> Cc: "engine-devel at ovirt.org" <devel at ovirt.org>
> >> Sent: Wednesday, January 21, 2015 6:05:51 PM
> >> Subject: [ovirt-devel] [DB] Diffrent UUIDs for inserted data per
> >> installation
> >> Hi
> >> I am trying to cleanup all the data insertion to the engine DB and make it
> >> more general
> >> The main drive to that is DB version squashing that was done manually and
> >> therefor was error prone ...
> >> I know that both storage_pool_id (a.k.a DC) and vds_group_id (a.k.a
> >> Cluster)
> >> needs to get a different UUID per installation.
> >> But I had found that UUIDs are generated per installation for also :
> >> table | column/s
> >> ------------------------------------
> >> [cpu_profiles] : id
> >> [gluster_services] : id
> >> [mac_pools] : id
> > Added Martin to comment on this one. AFAIK the id can be static for
> > all installations.
> Looks okay to have a static ID for the default MAC pool.
> >> [permissions] : id, object_id
> >> [vm_device]: device_id, vm_id
> >> [vm_static] : vm_guid
> >> [vnic_profiles] : id
> > Added Lior to comment. In any case, we can evaluate the id by querying
> > by name, hence the specific value id value shouldn't be important.
> We don't need to have distinct IDs by installation, however the current
> upgrade script creating the IDs (03_03_0710) creates one for each
> existing network, and every deployment will have its own networks... If
> your change might affect upgrades from versions < 3.3, then I don't see
> how you could achieve this using static IDs.
That's not a problem
Keep in mind that we are squashing always (latest-version - 3)
So, when we squash this 03_03_0710 script
1) upgrade for current installations will be supported only from 3.4 and up, that means that DB already run the 03_03_0710 script.
2) new installation , and then we have no problem either
The main point is that we squash only when we DON'T SUPPORT and upgrade from the squashed version, so he problem you had mentioned never occurs
> >> Please let me know if any of the above should be generated using the
> >> uuid_generate_v1() function on each installation or we can have static IDs
> >> for those.
> >> Thanks
> >> Eli Mesika
> >> _______________________________________________
> >> Devel mailing list
> >> Devel at ovirt.org
> >> http://lists.ovirt.org/mailman/listinfo/devel
More information about the Devel