| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-07-25 | |||
| 16:58:01 | dansmith | but we get that far only to find radios is None? | |
| 17:00:13 | mriedem | yeah | |
| 17:00:23 | mriedem | import error sets that rados package import to None | |
| 17:00:27 | mriedem | https://github.com/openstack/glance_store/blob/0.21.0/glance_store/_drivers/rbd.py#L41 | |
| 17:00:36 | mriedem | and rados isn't in the pip freeze output for py35 but it is for py27 | |
| 17:00:40 | mriedem | https://bugs.launchpad.net/glance-store/+bug/1706405 | |
| 17:00:41 | openstack | Launchpad bug 1706405 in glance_store "ceph jobs failing to upload images in pike due to "AttributeError: 'NoneType' object has no attribute 'Rados'" on py35" [Undecided,New] | |
| 17:00:49 | mriedem | so this is just some py35 package install thing for librados | |
| 17:01:00 | dansmith | no error log when we fail to import? | |
| 17:01:05 | mriedem | eff no | |
| 17:01:17 | mriedem | error logs are in the enterprise edition of glance-store | |
| 17:03:46 | gibi | mriedem: hi! Do you have time to joint to the notification subteam meeting or we should simply skip that? I'm not against skipping it today. | |
| 17:04:36 | sdague | mriedem: ok, how is your look on the experimental job? As I'm going to walk away from the computer for a bit shortly | |
| 17:05:22 | openstackgerrit | Merged openstack/nova master: [placement] Use wsgi_intercept in PlacementFixture https://review.openstack.org/486237 | |
| 17:06:11 | mriedem | sdague: going through results | |
| 17:06:15 | mriedem | but now in the notification meeting | |
| 17:06:21 | mriedem | so you might as well walk | |
| 17:12:44 | jangutter | OK, this is a python newbie question (due to lack of caffeine)... Why would vif.get('ovs_interfaceid',"foobar") return None? ( https://github.com/openstack/nova/blob/master/nova/network/os_vif_util.py#L281 ) | |
| 17:13:45 | openstackgerrit | Gábor Antal proposed openstack/nova master: Transform missing delete notifications https://review.openstack.org/410297 | |
| 17:15:05 | openstackgerrit | Gábor Antal proposed openstack/nova master: Send soft_delete from context manager https://review.openstack.org/476459 | |
| 17:15:12 | edleafe | jangutter: if there is a key named 'ovs_interfaceid' and its value is None | |
| 17:15:33 | edleafe | jangutter: it only returns the default param if the key isn't found | |
| 17:17:08 | mriedem | jangutter: so do vif.get('ovs_interfaceid',"foobar") or 'foobar' | |
| 17:17:33 | jangutter | edleafe: yeah, that makes sense. That's why I see the pattern there. | |
| 17:20:08 | sdague | lyarwood: was there any progress on this bug - https://bugs.launchpad.net/nova/+bug/1640506 ? | |
| 17:20:10 | openstack | Launchpad bug 1640506 in OpenStack Compute (nova) "Switch to os-brick encryptor provider implementations" [Wishlist,New] | |
| 17:20:53 | openstackgerrit | Chris Dent proposed openstack/nova master: Use wsgi-intercept in OSAPIFixture https://review.openstack.org/486825 | |
| 17:21:30 | sdague | mriedem: ok, brain sufficiently fried, so walking away from computer now, but the number of bugs in new is down to 60 - https://bugs.launchpad.net/nova/+bugs?search=Search&field.status=New - so it's back on one page now | |
| 17:22:09 | mriedem | yay | |
| 17:22:55 | openstackgerrit | Jan Gutter proposed openstack/nova master: Netronome SmartNIC Enablement https://review.openstack.org/483459 | |
| 17:23:02 | mriedem | dansmith: neutron dvr ha job is busted, but that's probably old news http://logs.openstack.org/56/477556/5/experimental/gate-tempest-dsvm-neutron-dvr-ha-multinode-full-ubuntu-xenial-nv/432c235/console.html#_2017-07-25_15_47_52_092840 | |
| 17:28:11 | jaypipes | mriedem, dansmith: if you have a chance to review https://review.openstack.org/#/c/483566/ that would be great. (the final placemnet-claims one) | |
| 17:28:38 | mriedem | jaypipes: yeah i was checking out the ci results before digging into the review | |
| 17:28:42 | jaypipes | k | |
| 17:28:43 | mriedem | should be done this afternoon | |
| 17:30:20 | openstackgerrit | Robert Ellis proposed openstack/nova master: Clarifying node_uuid usage in ironic driver. https://review.openstack.org/485803 | |
| 17:31:58 | openstackgerrit | Balazs Gibizer proposed openstack/nova master: Add 'updated_at' field to InstancePayload in notifications https://review.openstack.org/475276 | |
| 17:32:49 | gibi | mriedem: rebased the update_at bugfix ^^ | |
| 17:33:14 | gibi | mriedem: I mean I've rebased | |
| 17:33:54 | mriedem | thanks | |
| 17:42:01 | openstackgerrit | Mark Giles proposed openstack/nova master: Do not attempt volume swap when guest is stopped/suspended https://review.openstack.org/389798 | |
| 17:58:04 | openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova master: Remove the code related to extension loading from APIRouterV21 https://review.openstack.org/486414 | |
| 17:58:39 | openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova master: Remove the useless FakeExt https://review.openstack.org/486415 | |
| 17:58:47 | openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova master: Remove the useless extension block_device_mapping_v1 object https://review.openstack.org/486069 | |
| 17:59:45 | ildikov | mriedem: I added a comment to the translation patch | |
| 18:00:09 | ildikov | mriedem: I think the base for the confusion there is that the namings there are pretty confusing | |
| 18:00:31 | oomichi | alex_xu: re: https://review.openstack.org/#/c/486414/ yeah, that is an important one. +2 | |
| 18:00:37 | ildikov | mriedem: would that be fine to make that human readable or you want the structure change too? | |
| 18:02:23 | melwitt | sdague: yeah, that shouldn't be a thing with counting quotas in Pike. but maybe probably there needs to be a fix on stable only? I can't remember if we ever do that cc mriedem | |
| 18:02:27 | mriedem | ildikov: i said in https://review.openstack.org/#/c/486194/2/nova/volume/cinder.py@218 that we should rename that data_keys variable to connection_info | |
| 18:02:42 | mriedem | ildikov: however, _translate_attachment_ref doesn't return a connection_info dict | |
| 18:03:01 | mriedem | it mangles the attachment dict, | |
| 18:03:07 | mriedem | and adds a ['data'] key in it | |
| 18:03:17 | mriedem | and shoves the attachment['connection_info'] stuff in there | |
| 18:03:20 | mriedem | as far as i can tell | |
| 18:04:15 | ildikov | as connection_info is a free form data structure having stuff under the 'data' key and 'driver_volume_type' on top level with it basically fulfills the criteria as far as I can tell | |
| 18:05:21 | ildikov | but jgriffith is smarter than me on this front | |
| 18:06:06 | mriedem | ewww yeah i don't like that | |
| 18:06:10 | mriedem | and i just got now what it's doing | |
| 18:06:26 | mriedem | it's the attachment ref PLUS all of the crap from the connection_info, mainlined into the attachment ref body resp | |
| 18:06:35 | mriedem | that's super confusing and i don't think we should do that | |
| 18:06:49 | mriedem | let's just translate the connection_info within the attachment ref as i said in there | |
| 18:07:16 | ildikov | well, connection_info goes under 'data' as how it used to be before | |
| 18:07:32 | mriedem | thios https://review.openstack.org/#/c/486194/2/nova/volume/cinder.py@220 | |
| 18:07:37 | mriedem | *this | |
| 18:08:03 | mriedem | yes i get that | |
| 18:08:13 | mriedem | but it's also mangled into the attachment response body | |
| 18:09:04 | mriedem | if the cinder API returns attachment: {connection_info: {'foo': 'bar'}} and now we turn that into attachment: {'data': {'foo': 'bar'}} that gets confusing | |
| 18:10:01 | mriedem | i'd prefer to just see attachment: {connection_info: 'data': {{'foo': 'bar'}}} at the end | |
| 18:10:16 | mriedem | well ^ is busted, but you know | |
| 18:11:22 | jangutter | mriedem, jaypipes: https://review.openstack.org/#/c/486426 got the checkmark from Jenkins.... but did I throw my first exception correctly? | |
| 18:11:52 | mriedem | jangutter: omfg you don't throw anything! | |
| 18:12:02 | mriedem | :P | |
| 18:12:02 | jgriffith | mriedem ok, well this is why I dumped all that crap in the first place (which got us to that point) | |
| 18:12:04 | mriedem | you RAISE! | |
| 18:12:25 | jgriffith | mriedem the issue being that in my opinion that original return info was a confusing free form mess | |
| 18:12:39 | jangutter | mriedem: what the heck? is this Poker or Python? | |
| 18:12:43 | jgriffith | but then *we* decided we wanted to keep it consistent | |
| 18:12:44 | mriedem | jangutter: seems ok | |
| 18:12:58 | jgriffith | so that's why it's now stuffing in the way it is | |
| 18:13:38 | mriedem | jgriffith: i'm fine if cinder 3.27 wanted to flatten the connection_info dict and drop the 'data' subkey | |
| 18:14:01 | mriedem | but what i don't like is doing attachment_ref.update(attachment_ref.pop('connection_info', {})) basically | |
| 18:14:04 | mriedem | sans the driver_volume_type key | |
| 18:14:23 | mriedem | er attachment_ref.update(dict(data=attachment_ref.pop('connection_info', {}))) | |
| 18:14:47 | ildikov | mriedem: the removal of the 'data' key on the Nova side is quite an amount of code line change apparently :( | |
| 18:14:58 | mriedem | because then am i dealing with an attachment representation, or a connection_info, or some weird hybrid? | |
| 18:15:01 | ildikov | mriedem: that's why we thought to do that at another time | |
| 18:15:07 | mriedem | ildikov: yes it should be done another time | |
| 18:15:50 | jgriffith | mriedem so wait... do you have a better idea on how to translate that other than popping it out into a new struct? | |
| 18:15:58 | ildikov | mriedem: ok, at least one thing we agree at :) | |
| 18:18:33 | mriedem | jgriffith: i'm fine with popping out the original one and writing it back into attachment_ref['connection_info'] | |
| 18:18:42 | mriedem | i just don't want it munged into attachment_ref itself | |
| 18:18:53 | mriedem | per https://review.openstack.org/#/c/486194/2/nova/volume/cinder.py@220 | |
| 18:18:55 | jgriffith | mriedem that's reasonable | |
| 18:20:15 | jgriffith | mriedem ildikov I think I see the problem here... | |
| 18:20:41 | jgriffith | mriedem ildikov the attachment_create sadly returns a dict, while the attachment_update returns an attachment_ref object | |
| 18:21:10 | ildikov | jgriffith: don't we "play" with both? | |
| 18:21:33 | jgriffith | ildikov I don't know what you mean, but regardless... | |
| 18:21:39 | mriedem | jgriffith: yes the cinderclient code is confusing as well | |
| 18:21:55 | ildikov | jgriffith: that we translate both after calling to_dict() | |