Open Stack

Ihar, please see responses (inline).

-amrith

| -----Original Message-----
| From: Ihar Hrachyshka [mailto:ihrachys at redhat.com]
| Sent: Thursday, March 05, 2015 9:43 AM
| To: openstack-dev at lists.openstack.org
| Subject: Re: [openstack-dev] [Trove] request to backport the fix for bug
| 1333852 to juno
| 
| -----BEGIN PGP SIGNED MESSAGE-----
| Hash: SHA1
| 
| Not being involved in trove, but some general comments on backports.
| 
| On 03/04/2015 08:33 PM, Amrith Kumar wrote:
| > There has been a request to backport the fix for bug 1333852
| > (https://bugs.launchpad.net/trove/+bug/1333852) which was fixed in
| > Kilo into the Juno release.
| >
| 
| It would be easier if you directly link to patches in question.

[amrith] These are the patches that merged into Kilo
	https://review.openstack.org/#/c/115811/
	https://review.openstack.org/#/c/123301/

| 
| >
| >
| > The change includes a database change and a small change to the Trove
| > API. The change also requires a change to the trove client and the
| > trove controller code (trove-api). It is arguable whether this is a
| > backport or a new feature; I'm inclined to think it is more of an
| > extension of an existing feature than a new feature.
| >
| 
| It depends on what is a 'database change' above. If it's a schema change,
| then it's a complete no-go for backports. A change to API is also

[amrith] Yes, it is a schema change. See https://review.openstack.org/#/c/115811/

| suspicious, but without details it's hard to say. Finally, the need to
| patch a client to utilize the change probably means that it's not a bug
| fix (or at least, not an easy one).

[amrith] The change to the API is not that complex; it changes flavor from an int to a string and adds logic that knows how to tell one from the other.

| 
| Where do those flavor UUIDs come from? Were they present/supported in
| nova/juno?

[amrith] Yes. On a nova boot call for example, you could specify flavor-id thusly:

  --flavor <flavor>             Name or ID of flavor (see 'nova flavor-list').

With Trove (prior to this fix) you could only specify the ID which would be an integer.

| 
| >
| >
| > As such, I *don't* believe that this change should be considered a
| > good candidate for backport to Juno but I'm going to see whether there
| > is sufficient interest in this, to consider this change to be an
| > exception.
| >
| 
| Without details, it's hard to say for sure, but for initial look, the
| change you describe is too far stretching and has lots of issues that
| would make backport hard if not impossible.
|

[amrith] I agree. But I would appreciate input from others as well.
 
| /Ihar
| -----BEGIN PGP SIGNATURE-----
| Version: GnuPG v1
| 
| iQEcBAEBAgAGBQJU+GtwAAoJEC5aWaUY1u57+M4IAMjuF/f7OTMkaT1dxmy8GpV4
| /RoF06pPR5hU1oIjbjyvhRaqzTcKJBNqhuLzV7WhbynkyEuctg+QSqM/d2VQZwpp
| Gt59XEiIuLUYn46oC4J/S0DZBYHjRiZqcEXrJRozfzIvMQzqkCH+TeBxo9J5E/U4
| /I2rkGkDUm+XJa88M5PsTJP6Vp0nAvKQLa/Vjpe4/Ute2YMGlvFeH4NAsBy8XVWe
| BSJAIds0Abe1+uNwvaDeRbKaHwcgdAG/ia9WUO+8QHx1oXpLH/190o2P+xfZ8cno
| guPR2kSrzC0JLO5lfvRkjnDJd53kj/0tMf12xjzHHBC++grLUEs9i2AsvV/Dtyk=
| =s/sF
| -----END PGP SIGNATURE-----
| 
| __________________________________________________________________________
| OpenStack Development Mailing List (not for usage questions)
| Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
| http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev