Agree with removal of unsupported drivers. For supported drivers we will still need to support passthru. First, because it is currently used by many customers and for many drivers. Second, we will need to add support in Ironic and expose it in API for many features before we can do them without passthru crutch. For example, PXE NIC setup support, FW install/update, BIOS version setup/upgrade, etc.
Thanks, Arkady -----Original Message----- From: Mathieu Mitchell [mailto:[email protected]] Sent: Monday, November 14, 2016 10:42 AM To: OpenStack Development Mailing List (not for usage questions) <[email protected]> Subject: Re: [openstack-dev] [ironic] Removing agent vendor passthru and unsupported drivers Hi Pavlo, See my reply below. On 2016-11-14 7:50 AM, Pavlo Shchelokovskyy wrote: > Hi Ironicers, > > currently I'm busy with removing the lookup/heartbeats "as vendor passthru" > from Ironic which we slated for removal in Ocata, and have the > following question. > > Removing the old agent vendor passthru requires changes to some > unsupported drivers whose copies are already in > ironic-staging-drivers. The drivers in question are WoL, iBoot and > especially AMT (which uses a custom not-so-vendor passthru). The "follows-standard-deprecation" policy states the following "Features, APIs or configuration options are marked deprecated in the code. Appropriate warnings will be sent to the end user, operator or library user. **Code will be frozen and only receive minimal maintenance (just so that it continues to work as-is).**" [0] (emphasis mine). My understanding is that your changes would fall into the "just so that it continues to work as-is" clause. > > AFAIU according to our third-party drivers policy, those unsupported > drivers have to be removed from Ironic tree anyway (as there is no > plan to test them on third-party CI AFAIK) and this looks like a > perfect time to do it. > > So ideally I'd like to fix those in ironic-staging-drivers and then > remove them from Ironic tree via a depends-on patch. > > What do you think on such plan? The drivers were marked for removal in Ocata [1], so you can already remove them from the tree. A simple but relevant thing I note is that it would be preferable, from my point of view, to remove them all in a single commit. Finally, I would add that functional CI coverage for the SNMP driver is well under way [2]. We are currently doing the work to keep the SNMP driver in-tree (what we are doing is similar to VirtualBMC and the IPMI driver). Going ahead with a single commit to remove all the drivers would impact our current work. I would therefore suggest doing the required "vendor passthru" changes to the different drivers and post-pone the commit to delete all unsupported drivers. [0] https://governance.openstack.org/reference/tags/assert_follows-standard-deprecation.html#requirements [1] http://docs.openstack.org/releasenotes/ironic/current-series.html#id5 [2] https://review.openstack.org/#/q/status:open+topic:bug/1597793 Thank you, Mathieu Mitchell Internap > > Cheers, > Dr. Pavlo Shchelokovskyy > Senior Software Engineer > Mirantis Inc > www.mirantis.com > > > > ______________________________________________________________________ > ____ OpenStack Development Mailing List (not for usage questions) > Unsubscribe: > [email protected]?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
