*** jhfeng has quit IRC | 00:01 | |
*** woodster_ has quit IRC | 00:09 | |
*** su_zhang has quit IRC | 00:26 | |
*** mixos has joined #openstack-barbican | 00:35 | |
*** su_zhang has joined #openstack-barbican | 00:47 | |
*** spotz_zzz is now known as spotz | 00:51 | |
*** jaosorior has quit IRC | 00:53 | |
*** jaosorior has joined #openstack-barbican | 00:54 | |
*** gyee has quit IRC | 00:58 | |
*** jhfeng has joined #openstack-barbican | 01:04 | |
*** jhfeng has quit IRC | 01:06 | |
*** su_zhang has quit IRC | 01:07 | |
*** stevemar_ has quit IRC | 01:11 | |
*** ccneill_ has joined #openstack-barbican | 01:23 | |
*** ccneill has quit IRC | 01:25 | |
*** ccneill_ has quit IRC | 01:27 | |
*** spotz is now known as spotz_zzz | 01:35 | |
*** stevemar_ has joined #openstack-barbican | 02:00 | |
*** vivek-ebay has quit IRC | 02:05 | |
*** woodster_ has joined #openstack-barbican | 02:14 | |
*** vivek-ebay has joined #openstack-barbican | 03:16 | |
*** dave-mccowan has quit IRC | 03:16 | |
*** zz_dimtruck is now known as dimtruck | 03:40 | |
*** su_zhang has joined #openstack-barbican | 04:17 | |
*** woodster_ has quit IRC | 04:19 | |
*** vivek-ebay has quit IRC | 04:19 | |
*** jaosorior has quit IRC | 04:34 | |
*** jaosorior has joined #openstack-barbican | 04:34 | |
*** Nirupama has joined #openstack-barbican | 04:40 | |
*** everjeje has joined #openstack-barbican | 04:44 | |
*** mixos has quit IRC | 05:22 | |
*** dimtruck is now known as zz_dimtruck | 05:24 | |
*** jamielennox is now known as jamielennox|away | 05:30 | |
*** stevemar_ has quit IRC | 05:31 | |
*** su_zhang has quit IRC | 05:46 | |
*** edtubill has joined #openstack-barbican | 06:08 | |
*** jamielennox|away is now known as jamielennox | 06:21 | |
*** shohel has joined #openstack-barbican | 06:25 | |
*** edtubill has quit IRC | 06:42 | |
*** jaosorior has quit IRC | 08:57 | |
*** jaosorior has joined #openstack-barbican | 08:58 | |
*** everjeje has quit IRC | 09:16 | |
*** darrenmoffat has quit IRC | 09:21 | |
*** darrenmoffat has joined #openstack-barbican | 09:22 | |
*** jaosorior has quit IRC | 09:55 | |
*** jaosorior has joined #openstack-barbican | 09:55 | |
*** mmdurrant has quit IRC | 10:09 | |
*** jaosorior has quit IRC | 10:33 | |
*** jaosorior has joined #openstack-barbican | 10:34 | |
*** dave-mccowan has joined #openstack-barbican | 10:54 | |
jaosorior | dave-mccowan: Hey man, good morning | 11:00 |
---|---|---|
dave-mccowan | good morning ozz | 11:00 |
jaosorior | how's it going? | 11:00 |
dave-mccowan | it's going well.... the quiet between releases; good chance to catch up on stuff i've been putting off. | 11:01 |
jaosorior | haha yeah | 11:01 |
jaosorior | anything of yours that I've missed reviewing? | 11:02 |
jaosorior | dave-mccowan: Waiting for a long deployment to finish, so I've got plenty of time for reviewing :P | 11:04 |
dave-mccowan | nope. but, maybe one CR we could think about. https://review.openstack.org/#/c/171023/ In this CR, the request was that "validation stuff" should stop leaking out of the validator. since then, this problem has gotten worse. as we add new features with per-project aspects, we're adding validation checks for ownership in the controllers. (cas.py got some checks for that this cycle) do you have any fresh ideas there, or should we | 11:06 |
dave-mccowan | give up? | 11:06 |
jaosorior | dave-mccowan: good that you mention it. Let me finish up a CR and I'll take a better look into that. | 11:07 |
*** jamielennox is now known as jamielennox|away | 11:19 | |
jaosorior | dave-mccowan: No fresh ideas really. We should start doing some refactoring to fix things... I, however, want the validation classes to be more precise in what they do | 11:27 |
jaosorior | so perhaps we should separate between schema validation, and other types of validation | 11:27 |
dave-mccowan | jaosorior centralizing ownership validation sounds like a good idea. it would be a good reminder for developers to make sure those checks are implemented for new features too. | 11:37 |
jaosorior | dave-mccowan: exactly | 11:38 |
dave-mccowan | jaosorior i'm not sure there is anything useful in the current patch. i think i'll abandon it and open a wishlist bug. thoughts? | 11:40 |
jaosorior | +1 to that | 11:41 |
*** mmdurrant has joined #openstack-barbican | 11:58 | |
*** Nirupama has quit IRC | 12:01 | |
*** david-ly_ has joined #openstack-barbican | 12:07 | |
*** david-lyle has quit IRC | 12:09 | |
*** david-ly_ is now known as david-lyle | 12:09 | |
*** su_zhang has joined #openstack-barbican | 12:23 | |
*** su_zhang has quit IRC | 12:56 | |
dave-mccowan | alee ping | 13:15 |
*** su_zhang has joined #openstack-barbican | 13:16 | |
*** david-lyle has quit IRC | 13:16 | |
dave-mccowan | alee i noticed the discussion under this CR talking about why Nova get's 404 when trying to get a key. https://review.openstack.org/#/c/211114/1. looks like the work-around (or fix) is to add /v1 to the URI in the service catalog. | 13:16 |
alee | dave-mccowan, lookiing ,, | 13:17 |
alee | dave-mccowan, interesting .. let me take a look at my logs again .. | 13:20 |
*** jaosorior has quit IRC | 13:31 | |
*** jaosorior has joined #openstack-barbican | 13:32 | |
*** david-lyle has joined #openstack-barbican | 13:32 | |
alee | dave-mccowan, that doesn't appear to have helped | 13:50 |
*** spotz_zzz is now known as spotz | 13:51 | |
dave-mccowan | alee can you tell if Nova is using the right URI? the comments imply that it is building it's own using only the secret id and catalog. | 13:51 |
alee | dave-mccowan, I think you might be onto something --trying again .. | 13:54 |
*** jaosorior has quit IRC | 13:55 | |
*** jaosorior has joined #openstack-barbican | 13:55 | |
alee | dave-mccowan, yes - you're right -- it looks like its not getting the right url | 13:56 |
alee | dave-mccowan, I see the request in the nova-compute.log .. | 13:56 |
alee | let me try with that auth_token on the command line | 13:56 |
alee | dave-mccowan, yes - its definitely not using the right url | 13:58 |
alee | ie. not adding the v1 -- good catch | 13:58 |
alee | dave-mccowan, but I think the endpoint in the service catalog now has v1 in i t.. | 13:59 |
dave-mccowan | alee i think there are two entries in the service catalog. the admin and the non-admin. did you catch both? | 14:00 |
alee | dave-mccowan, yup | 14:00 |
*** spotz is now known as spotz_zzz | 14:01 | |
*** spotz_zzz is now known as spotz | 14:01 | |
alee | dave-mccowan, there are some parameters in nova.conf that might help -- not sure how to use them .. | 14:05 |
alee | #endpoint_template=<None> | 14:05 |
*** jhfeng has joined #openstack-barbican | 14:06 | |
dave-mccowan | LOL https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L105 | 14:09 |
dave-mccowan | alee i think we need a fix to nova to get this work. | 14:10 |
alee | dave-mccowan, I'm not sure -- maybe the versioning should be handled within the barbican client | 14:12 |
alee | dave-mccowan, redrobot - should adding the '/v1' be something that is handled y the client? | 14:15 |
alee | barbican-client? | 14:16 |
dave-mccowan | alee that goes back to the philosophical debate in the CR. it's definitely easier for us to fix in barbican-client, then get Nova to change. should be an easy hack to get the client to change an un-versioned request to /v1/. that will work until /v2/. :-) | 14:16 |
alee | dave-mccowan, I'm just surprised this is broken .. | 14:16 |
alee | (and no one cotttoned onto it) | 14:17 |
alee | are there no functional tests that test this? | 14:17 |
*** xaeth_afk is now known as xaeth | 14:18 | |
dave-mccowan | alee i think cross-project gate tests are rare :-( | 14:18 |
alee | I'd like to see if kfarr, rellerreller have this working | 14:18 |
alee | dave-mccowan, trying the endpoint_template to see if I can work around it for now .. | 14:19 |
dave-mccowan | is the key being ordered? or pre-stored? | 14:19 |
dave-mccowan | alee looking at the nova code with the link above, the template is parsed at 99 and /v1/ removed at 105. | 14:20 |
dave-mccowan | alee ah... maybe put /v1/v1/ in the catalog. :-/ | 14:20 |
alee | dave-mccowan, well I can put /v1/v1 in the template | 14:21 |
alee | dave-mccowan, we should look at what cinder does | 14:21 |
alee | dave-mccowan, because they do it correctly | 14:21 |
alee | and the request goes through correctly there | 14:21 |
*** kfarr has joined #openstack-barbican | 14:21 | |
alee | kfarr, ! | 14:22 |
kfarr | alee! Hi! I got in late today and am still getting things set up | 14:22 |
alee | kfarr, no worries. dave-mccowan has pinpointed what the problem is .. | 14:22 |
alee | kfarr, and now looking for a solution .. | 14:23 |
alee | kfarr, when nova goes to get the key , it constructs a url without the v1 in it | 14:23 |
kfarr | alee oh ok, I know the fix for that, one sec | 14:23 |
dave-mccowan | kfarr check out our discussion here. you reviewed the patch we were looking at a couple months ago. http://eavesdrop.openstack.org/irclogs/%23openstack-barbican/%23openstack-barbican.2015-10-07.log.html | 14:24 |
* dave-mccowan places bet kfarr comes back with /v1/v1/ :-) | 14:25 | |
alee | dave-mccowan, I'm using an old client-- let me try updating .. | 14:25 |
kfarr | in nova.conf, put an entry for endpoint_template | 14:26 |
kfarr | http://localhost:9311/v1/%(project_id)s | 14:26 |
kfarr | This issue is fixed in Castellan, bte | 14:26 |
kfarr | btw | 14:26 |
dave-mccowan | kfarr, alee. interesting. this code will strip off the {project_id} and leave /v1/, so it works. https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L105 | 14:28 |
*** xaeth is now known as xaeth_afk | 14:28 | |
alee | dave-mccowan, kfarr -- juts updated my barbican-client , and now it works I think .. | 14:28 |
alee | so it seems the fix was in barbican client | 14:29 |
kfarr | alee, what was the fix? | 14:31 |
kfarr | I really think you only need to specify the endpoint_template in nova.conf | 14:31 |
alee | kfarr, I think specifying the endpoint_template in nova.conf will work | 14:31 |
alee | kfarr, on the other hand, it seems that updating my barbican-client to the latest or similar appears to work too. | 14:32 |
alee | testing to confirm but it appears that the key was retrieved | 14:32 |
alee | kfarr, dave-mccowan ie. using latest client means not having to specify endpoint_template | 14:33 |
alee | of course, I now have /v1 in the catalog now .. let me remove that .. | 14:33 |
alee | not that that shoudl make a difference I think due to nova code dave-mccowan pointed out | 14:34 |
dave-mccowan | alee, kfarr, i don't see any code in the client that allows for /v1/ and not /v1/. are you sure you don't also have a change to endpoint template? | 14:34 |
kfarr | Ok, so in that code snippet in nova, it strips the v1, but only for when it passes the uri to the barbican client | 14:35 |
kfarr | the real issue is that when that code was written, barbican had the v1 in the service catalog | 14:35 |
kfarr | then somewhere along the way, that changed because keystone is now recommending removing the v1 | 14:35 |
kfarr | so the code doesn't work unless you override the endpoint to have a v1 | 14:36 |
dave-mccowan | kfarr and you have to override with http://localhost:9311:/v1/foo, because line 105 strips off the end of the URI. | 14:37 |
kfarr | Yes that line strips off the v1, but that URI with the v1 stripped off is only used to create the barbican client | 14:39 |
kfarr | The problem is when it creates the secret_ref | 14:39 |
*** shohel has quit IRC | 14:39 | |
kfarr | because it manually builds the secret ref given a uuid | 14:39 |
kfarr | https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L233 | 14:39 |
*** xaeth_afk is now known as xaeth | 14:39 | |
kfarr | it uses: self._base_url + "/secrets/" + key_id | 14:40 |
kfarr | Back in the __init__ function, if you don't provide an endpoint template, it goes to the endpoint catalog | 14:40 |
kfarr | and the barbican uri in the endpoint catalog doesn't have a v1 | 14:41 |
kfarr | self._base_url = _SESSION.get_endpoint(...) | 14:42 |
dave-mccowan | kfarr gotcha. _base_url is not stripped of it last param. which make me wonder, why do you have %(project_id)s in your example? | 14:42 |
kfarr | because it does string formatting: | 14:42 |
kfarr | https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L99 | 14:42 |
dave-mccowan | kfarr is project id part of the secret ref? | 14:43 |
kfarr | dave-mccowan, good point | 14:44 |
kfarr | dave-mccown, then maybe you can just override it to this: http://localhost:9311/v1/ | 14:45 |
*** zz_dimtruck is now known as dimtruck | 14:45 | |
alee | dave-mccowan, kfarr - is there a better way to test http://docs.openstack.org/juno/config-reference/content/section_testing_encryption.html ? | 14:45 |
alee | dave-mccowan, kfarr - the test will show the string being written to the unencrypted volume just fine. | 14:46 |
alee | and that we cant find the string in the encrypted volume | 14:46 |
alee | how is that different from not having written to the encryoted volume at all? | 14:47 |
*** silos has joined #openstack-barbican | 14:48 | |
kfarr | alee, that's always how we demoed it | 14:48 |
alee | kfarr, yup - not criticising the test -- it makes sense. just wondering if we can do it in a way that proves we actually wrote the partition | 14:49 |
*** lvh has quit IRC | 14:49 | |
alee | kfarr, maybe I'll try mounting the disk .. | 14:50 |
dave-mccowan | can you cat something before and after and show it growing? also read and write, the value must have come back from somewhere. | 14:50 |
*** stevemar_ has joined #openstack-barbican | 14:51 | |
alee | yeah -- maybe read .. | 14:51 |
*** lvh has joined #openstack-barbican | 14:51 | |
kfarr | alee, just to clarify, were you able to get it working by modifying the endpoint_template or did you have to fiddle with other things too? | 14:52 |
dave-mccowan | alee i assume if you can grep "/dev/stack-volumes/volume-*", you can also cat them. | 14:52 |
alee | kfarr, give me just a sec -- I had added v1 to the catalog .. let me remove and connnnnnnnfirm | 14:54 |
*** kebray has joined #openstack-barbican | 14:57 | |
*** kebray has quit IRC | 14:58 | |
*** edtubill has joined #openstack-barbican | 14:58 | |
*** kebray has joined #openstack-barbican | 14:59 | |
*** diazjf has joined #openstack-barbican | 15:00 | |
*** diazjf has left #openstack-barbican | 15:01 | |
*** mixos has joined #openstack-barbican | 15:02 | |
alee | kfarr, but to be clear -- I updated my python client and then things started to work | 15:05 |
alee | no template change needed | 15:05 |
kfarr | alee do you know what is different about the new python client? | 15:05 |
*** jaosorior has quit IRC | 15:06 | |
alee | kfarr, presumably it handles the versions? have not looked yet .. | 15:06 |
*** jaosorior has joined #openstack-barbican | 15:06 | |
alee | kfarr, still confirming | 15:06 |
*** kebray has quit IRC | 15:14 | |
*** everjeje has joined #openstack-barbican | 15:17 | |
*** alee_ has joined #openstack-barbican | 15:19 | |
alee | kfarr, dave-mccowan ok definitely works with new python-barbicanclient | 15:24 |
kfarr | alee, great! | 15:24 |
alee | kfarr, no special nova config required | 15:24 |
* dave-mccowan wonders how o.O | 15:25 | |
alee | kfarr, dave-mccowan -- I also added a step where I cat the disk and grep for the phrase I wrote while on the vm | 15:25 |
alee | works on the vm -- doesn't work outside | 15:26 |
alee | as you'd expect | 15:26 |
dave-mccowan | alee sweet. are you going to push some doc or wiki? | 15:26 |
jaosorior | alee: What works with the new python-barbicanclient? | 15:26 |
jaosorior | anybody knows where redrobot is? | 15:27 |
jaosorior | People have been asking for a release of python-barbicanclient | 15:27 |
alee | dave-mccowan, better than that -- I'm working on a fork of some scripts to put it all together | 15:27 |
*** xaeth is now known as xaeth_afk | 15:28 | |
alee | dave-mccowan, the scripts will set up two vms - one with ipa (including dogtag) and one with barbican/ packstack | 15:28 |
alee | and will set up and test volume encryption | 15:28 |
alee | dave-mccowan, this is for the tokyo demo. | 15:29 |
dave-mccowan | alee awesome. where will you run the demo? do you have a presentation? | 15:29 |
alee | dave-mccowan, I'll also do silly things like register barbican as a service in ipa so we can get an ssl cert so we have https running on barbican | 15:30 |
*** kebray has joined #openstack-barbican | 15:30 | |
alee | dave-mccowan, red hat booth | 15:30 |
alee | dave-mccowan, there will be a video .. | 15:30 |
alee | dave-mccowan, now that its working I can start putting it together | 15:31 |
alee | dave-mccowan, I'll sned you the link to the repo once I update it | 15:31 |
alee | jaosorior, I was using an old pythonclient and retrieving the key rom nova was failing with 404 | 15:32 |
dave-mccowan | alee thanks. any clue on where in the code the /v1/ is added back? i'm going to lose sleep. :-/ :-) | 15:32 |
jaosorior | alee, ooh, I see | 15:32 |
alee | jaosorior, the reason for that was that nova was constructing a url without the v1 | 15:32 |
alee | jaosorior, I updated my client and now it automagically works | 15:33 |
jaosorior | alee: yeah, we did some changes to how the endpoints are handled | 15:33 |
alee | jaosorior, so sometime someone fixed the client to make them handle the endpoint and add back the v1 | 15:33 |
alee | jaosorior, maybe even you :) | 15:33 |
jaosorior | it was me actually haha | 15:33 |
jaosorior | IIRC, it's been a while | 15:34 |
jaosorior | which makes me think | 15:34 |
jaosorior | damn, you must have had a pretty old client | 15:34 |
alee | jaosorior, well - the client has only been updated in fedora recently | 15:34 |
jaosorior | alee, I see | 15:35 |
jaosorior | but yeah, we need to poke redrobot about a new python-barbicanclient release | 15:35 |
dave-mccowan | jaosorior do you have a link to the CR? i want to add it as a comment to another CR, for people who run across this issue. | 15:35 |
alee | jaosorior, and even then it was later than that -- I think I had version 3.2.X | 15:35 |
*** kebray has quit IRC | 15:35 | |
alee | jaosorior, and no I have version 3.3.x | 15:35 |
alee | jaosorior, so 3.2.X to 3.3.0 | 15:36 |
alee | jaosorior, that doesn't sound that long ago .. | 15:37 |
jaosorior | alee: Then it might been another issue than what I'm thinking | 15:37 |
jaosorior | what I had done is enable the getting of the endpoint through the keystone catalog, which we weren't doing before | 15:38 |
alee | jaosorior, right -- but the keystone catalog entry used to include a v1 in it -- it no longer does | 15:39 |
alee | jaosorior, so code must have been added between 3.2 and 3.3 to add it back | 15:39 |
*** kebray has joined #openstack-barbican | 15:39 | |
alee | (or handle it properly) | 15:40 |
jaosorior | dave-mccowan: https://review.openstack.org/#/c/195453/ | 15:40 |
*** david-lyle has quit IRC | 15:41 | |
*** kebray has quit IRC | 15:42 | |
*** david-lyle has joined #openstack-barbican | 15:42 | |
*** vivek-ebay has joined #openstack-barbican | 15:43 | |
*** arunkant has quit IRC | 15:48 | |
*** david-ly_ has joined #openstack-barbican | 15:55 | |
*** david-lyle has quit IRC | 15:55 | |
*** david-ly_ has quit IRC | 15:56 | |
*** david-lyle has joined #openstack-barbican | 15:56 | |
openstackgerrit | Douglas Mendizábal proposed openstack/barbican: Add RBAC docs for Cloud Administrator Guide https://review.openstack.org/231222 | 15:57 |
*** lisaclark1 has joined #openstack-barbican | 16:01 | |
*** lisaclark1 has quit IRC | 16:02 | |
*** arunkant has joined #openstack-barbican | 16:03 | |
*** lisaclark1 has joined #openstack-barbican | 16:04 | |
*** xaeth_afk is now known as xaeth | 16:07 | |
*** kebray has joined #openstack-barbican | 16:08 | |
lisaclark1 | morning barbicaneers | 16:11 |
lisaclark1 | does anyone know if there is a summary of all blueprints / bug fixes in the Liberty release? or must I review the launchpads of our 3 milestone releases and the 2 RCs to get this summary view? | 16:12 |
*** su_zhang has quit IRC | 16:14 | |
redrobot | mornin' lisaclark1 | 16:18 |
redrobot | lisaclark1 I don't think there's such a page yet. Once RC2 becomes Liberty 1.0.0 then the release page will have all that info. | 16:19 |
redrobot | lisaclark1 just like https://launchpad.net/barbican/kilo/2015.1.0 for Kilo | 16:20 |
jaosorior | redrobot: Any python-barbicanclient release coming soon? :D | 16:21 |
redrobot | jaosorior haven't taken a look recently, but we're definitely due for one | 16:21 |
redrobot | jaosorior how much can we still change the openstack cli plugin after release? | 16:22 |
*** spotz is now known as spotz_zzz | 16:23 | |
lisaclark1 | thanks redrobot. that's the page for Liberty that I want ;-) | 16:29 |
arunkant | dave-mccowan: ping | 16:38 |
dave-mccowan | arunkant pong | 16:38 |
arunkant | dave-mccowan: Just now saw your comment on https://review.openstack.org/#/c/211114/ | 16:38 |
arunkant | you mentioned its fixed by another review. But the issue is on nova side and not on barbicanclient | 16:39 |
dave-mccowan | alee,kfarr, jaosorior, and i have been talking about it this morning. alee got his deployment working by upgrading to a client with jaosorior's fix. | 16:40 |
arunkant | dave-mccowan: Was he testing nova emphermal or cinder volume encryption..cinder volume encryption works in devstack because default encryption_api_url has /v1 in it | 16:42 |
*** vivek-ebay has quit IRC | 16:42 | |
arunkant | dave-mccowan: whereas nova ephemeral tries to use keystoneclient to leverage version discovery which does not work | 16:44 |
dave-mccowan | alee ^^ arunkant from code inspection, i agreed with you. i did not expect it to work. but, it did for alee. | 16:45 |
arunkant | dave-mccowan: we tested nova emphermal part last week and it works when specify /v1 in devstack local conf or in barbican endpoint in service catalog.. | 16:47 |
arunkant | alee: Were you testing nova emphermal storage or cinder volume encryption ? | 16:48 |
alee | arunkant, testing cinder volume encryption -- and using packstack | 16:48 |
alee | arunkant, where is the encryption_api_url set? | 16:49 |
arunkant | alee, okay..yes..that works because https://github.com/openstack/cinder/blob/master/cinder/keymgr/key_mgr.py#L29 | 16:49 |
arunkant | alee, so default URL has /v1 in it and it works in devstack without any change. | 16:50 |
alee | arunkant, ok yes -- have not had any issues with cinder | 16:50 |
alee | arunkant, I had issues with nova until I unpgraded my client | 16:50 |
*** xaeth is now known as xaeth_afk | 16:51 | |
arunkant | alee, so it works as that URL is used as-is when creating secret_ref in cinder side..https://github.com/openstack/cinder/blob/master/cinder/keymgr/barbican.py#L46 and https://github.com/openstack/cinder/blob/master/cinder/keymgr/barbican.py#L217 | 16:52 |
*** lisaclark1 has quit IRC | 16:52 | |
alee | arunkant, why are we creating the url at all? don't we return a reference to the secret in the order? | 16:54 |
arunkant | alee, but in nova emphermal storage case..it does not as there is no default defined and they try to identify url using keystoneclient version discovery..https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L102 | 16:54 |
arunkant | alee, I think..service client stores only secret uuid .. | 16:54 |
alee | arunkant, perhaps that should be changed -- this code will not work when we have barbican federation | 16:55 |
arunkant | alee, so they need to construct secret_ref URL when they need to read the secret.. | 16:55 |
arunkant | alee, yes it needs to be changed to provide "version" as additional parameter for version discovery | 16:56 |
alee | arunkant, so to be clear .. | 16:57 |
arunkant | So may be add additional input in https://github.com/openstack/nova/blob/master/nova/keymgr/barbican.py#L93 | 16:57 |
*** kebray has quit IRC | 16:57 | |
alee | arunkant, when i was cinder volume encryption with old barbican client, I was able to order and store the key in cinder, but was not able to retrive the key from nova to attach it to a vm | 16:58 |
alee | arunkant, because nova was not adding the v1 | 16:58 |
alee | arunkant, I would think it goes through the code you mention above. | 16:59 |
*** silos has left #openstack-barbican | 16:59 | |
alee | arunkant, with the new python-barbicanclient, it just works | 16:59 |
alee | arunkant, do you have a link so I can test nova ephemeral? | 17:00 |
arunkant | For cinder volume encryption.. For us, we tested like 4 weeks or may be earlier..it was working in devstack with default config | 17:00 |
arunkant | alee, its quite similar to cinder volume..but only thing is in devstack..you will need to set encryption_api_url , with /v1, in addition to setting keymt api_class | 17:03 |
arunkant | s/keymt/ keymgr | 17:03 |
openstackgerrit | Christopher Solis proposed openstack/barbican: Update Devstack documentation https://review.openstack.org/230276 | 17:03 |
*** lisaclark1 has joined #openstack-barbican | 17:03 | |
alee | arunkant, and then test in exactly the same way? | 17:04 |
arunkant | alee, this was the devstack conf ..http://paste.openstack.org/show/475637/ | 17:06 |
*** su_zhang has joined #openstack-barbican | 17:11 | |
arunkant | alee, I agree keymgr code on nova ephemeral storage should be modified to leverage version discovery and may be on cinder side, version discovery support needs to be added. | 17:11 |
alee | arunkant, I'll try it out | 17:12 |
*** vivek-ebay has joined #openstack-barbican | 17:13 | |
*** spotz_zzz is now known as spotz | 17:15 | |
kfarr | arunkant alee in castellan, version discovery was implemented. Hopefully in the next cycle we will be working on replacing the nova and cinder key managers with castellan | 17:24 |
arunkant | kfarr : +1 | 17:26 |
kfarr | Though there still is a possible problem if federated Barbican is implemented because Castellan also does the manual recreation of secret refs | 17:29 |
*** xaeth_afk is now known as xaeth | 17:44 | |
lisaclark1 | hi barbicaneers: quick poll of the room. anyone happen to have a summit registration ticket that they're not able to use? | 17:44 |
kfarr | lisaclark1 I have a registration code that I will not be using! I will only be there in spirit :( | 17:47 |
lisaclark1 | kfarr: sorry to hear you won't be there :-(. did you register with your code already and have an eventbrite ticket? | 17:48 |
kfarr | lisaclark1, no I do not have an eventbrite ticket | 17:48 |
lisaclark1 | thanks kfarr. i have a code also, but at this date it gives you a discount off the ticket price, but not a $0 ticket. i was hoping to find someone that did register their code but isn't using their ticket. | 17:49 |
kfarr | lisaclark1, oh I see! | 17:50 |
*** kebray has joined #openstack-barbican | 17:59 | |
*** xek has quit IRC | 18:07 | |
*** xek has joined #openstack-barbican | 18:08 | |
*** su_zhang has quit IRC | 18:14 | |
*** xaeth is now known as xaeth_afk | 18:22 | |
*** silos has joined #openstack-barbican | 18:26 | |
openstackgerrit | Christopher Solis proposed openstack/barbican: Update Devstack documentation https://review.openstack.org/230276 | 18:29 |
*** xaeth_afk is now known as xaeth | 18:29 | |
*** su_zhang has joined #openstack-barbican | 18:34 | |
arunkant | can a core look into this and possibly provide workflow..https://review.openstack.org/#/c/208344/ | 18:43 |
*** lisaclark1 has quit IRC | 18:43 | |
*** jaosorior has quit IRC | 18:47 | |
*** jaosorior has joined #openstack-barbican | 18:47 | |
jaosorior | redrobot ping | 18:47 |
redrobot | jaosorior pong | 18:47 |
*** kebray has quit IRC | 18:58 | |
*** kebray has joined #openstack-barbican | 19:01 | |
*** su_zhang has quit IRC | 19:02 | |
*** kebray has quit IRC | 19:04 | |
*** diazjf has joined #openstack-barbican | 19:12 | |
*** diazjf has left #openstack-barbican | 19:14 | |
*** lisaclark1 has joined #openstack-barbican | 19:19 | |
arunkant | kfarr: ping . | 19:20 |
kfarr | arunkant pong! | 19:21 |
arunkant | kfarr: Hi..question about kmip plugin. Is it possible to specify multiple host in kmip server host ? | 19:22 |
kfarr | arunkant, no, it only supports one host | 19:22 |
kfarr | arunkant, I think that was how the talk of federated Barbican got started, because each Barbican can only have one backend, including KMIP | 19:23 |
arunkant | kfarr: Oh..so how are deployments supposed to provide HA around KMIP servers ? | 19:24 |
arunkant | kfarr: One backend is fine..but the question is more around having multiple KMIP servers containing same data..primarily for high availability | 19:25 |
*** nelsnels_ has joined #openstack-barbican | 19:25 | |
*** nelsnelson has quit IRC | 19:25 | |
*** ryanpetrello has quit IRC | 19:27 | |
kfarr | arunkant, oh ok. I did not realize that any of the other backends had failover options? The current design meets our needs so far, but you pose an interesting point | 19:27 |
*** ryanpetrello has joined #openstack-barbican | 19:27 | |
arunkant | kfarr, in some clients library, client can switch to different server if the one of them happens to be down/unreachable for some reason. Otherwise client application has to implement that logic. | 19:30 |
*** lisaclark1 has quit IRC | 19:31 | |
*** su_zhang has joined #openstack-barbican | 19:33 | |
*** jaosorior has quit IRC | 19:36 | |
*** su_zhang has quit IRC | 19:37 | |
arunkant | alee, in barbican, does dogtag plugin supports multiple host for client connection ? | 19:41 |
alee | arunkant, not sure I understand what you mean by that ? | 19:42 |
alee | arunkant, can I connect from where to where ? and what is multiple? | 19:43 |
openstackgerrit | Merged openstack/barbican: Updated from global requirements https://review.openstack.org/231224 | 19:43 |
arunkant | alee, the question is around having multiple dogtag servers for HA ..does plugin supports that ? | 19:43 |
alee | arunkant, gotcha -- so right now -- dogtag plugin can only talk to a single dogtag ca .. but .. | 19:44 |
alee | dogtag has the ability to clone cas. | 19:45 |
alee | so you basically end up with another ca that has the same signing certs and keys -- to all extents and purposes the same ca as the original | 19:45 |
arunkant | alee, okay..so I am guessing cloning means creating passive server with same data.. | 19:46 |
alee | with data replicated between them using the underlyting db | 19:46 |
alee | arunkant, they can be active active | 19:46 |
alee | as they issue certs within different serial number ranges | 19:46 |
alee | and the data is replicated | 19:46 |
*** everjeje has quit IRC | 19:46 | |
*** dhellmann has quit IRC | 19:47 | |
alee | so if you had such a scenario - then you could put a load balancer in fron ofthe cas | 19:47 |
*** dhellmann has joined #openstack-barbican | 19:47 | |
alee | and configure the plugin to talk to the vip on the load baklancer | 19:47 |
arunkant | alee, okay...what will be the process to make barbican use that cloned server in case primary went down for some reason ? | 19:47 |
alee | well - if its a load balancer and one server is down, the vip will automatically direct all traffic to the other server | 19:48 |
alee | incidentally both cas and kras are cloned | 19:49 |
arunkant | alee, okay...so haproxy (LB) kind of solution can handle dogtag session and request offloading to available server ? | 19:50 |
alee | yes | 19:50 |
openstackgerrit | Merged openstack/barbican: Add RBAC docs for Cloud Administrator Guide https://review.openstack.org/231222 | 19:51 |
alee | arunkant, this is how dogtag customers handle HA and load balancing for just dogtag deployments | 19:51 |
arunkant | alee, okay. great. Thanks for clarifying it. | 19:51 |
alee | arunkant, np -- let me know if you're trying to set it up :) | 19:52 |
arunkant | alee, will reach out to you guys in near future..trying to understand how barbican plugin servers HA is handled. | 19:53 |
*** xaeth is now known as xaeth_afk | 19:55 | |
*** lisaclark1 has joined #openstack-barbican | 20:16 | |
*** kebray has joined #openstack-barbican | 20:17 | |
*** kfarr has quit IRC | 20:26 | |
*** xaeth_afk is now known as xaeth | 20:27 | |
*** xaeth is now known as xaeth_afk | 20:39 | |
*** xaeth_afk is now known as xaeth | 20:46 | |
*** su_zhang has joined #openstack-barbican | 20:52 | |
*** atiwari1 has quit IRC | 20:53 | |
*** spotz is now known as spotz_zzz | 21:03 | |
*** mixos has quit IRC | 21:07 | |
*** silos has left #openstack-barbican | 21:08 | |
*** lisaclark1 has quit IRC | 21:27 | |
*** jamielennox|away is now known as jamielennox | 21:32 | |
*** edtubill has quit IRC | 21:36 | |
*** diazjf has joined #openstack-barbican | 21:40 | |
*** diazjf has quit IRC | 21:40 | |
*** spotz_zzz is now known as spotz | 21:48 | |
*** kebray has quit IRC | 21:48 | |
*** xaeth is now known as xaeth_afk | 21:52 | |
*** spotz is now known as spotz_zzz | 21:57 | |
*** su_zhang_ has joined #openstack-barbican | 22:07 | |
*** lisaclark1 has joined #openstack-barbican | 22:07 | |
*** lisaclark1 has quit IRC | 22:08 | |
*** su_zhang has quit IRC | 22:11 | |
*** lisaclark1 has joined #openstack-barbican | 22:12 | |
*** kebray has joined #openstack-barbican | 22:17 | |
jhfeng | have anyone tried using softHSM with Barbican in devstack ? | 22:24 |
*** lisaclark1 has quit IRC | 22:35 | |
*** stevemar_ has quit IRC | 22:36 | |
*** stevemar_ has joined #openstack-barbican | 22:37 | |
*** lisaclark1 has joined #openstack-barbican | 22:38 | |
*** dimtruck is now known as zz_dimtruck | 22:41 | |
*** stevemar_ has quit IRC | 22:41 | |
*** su_zhang has joined #openstack-barbican | 22:49 | |
*** su_zhang_ has quit IRC | 22:52 | |
*** david-lyle has quit IRC | 22:54 | |
*** david-lyle has joined #openstack-barbican | 22:55 | |
*** lisaclark1 has quit IRC | 22:57 | |
*** lisaclark1 has joined #openstack-barbican | 22:57 | |
*** stevemar_ has joined #openstack-barbican | 22:59 | |
*** lisaclark1 has quit IRC | 23:01 | |
*** jhfeng has quit IRC | 23:06 | |
*** david-lyle has quit IRC | 23:10 | |
*** david-lyle has joined #openstack-barbican | 23:10 | |
*** david-lyle has quit IRC | 23:14 | |
*** yuanying has quit IRC | 23:16 | |
*** david-lyle has joined #openstack-barbican | 23:17 | |
*** david-ly_ has joined #openstack-barbican | 23:19 | |
*** yuanying has joined #openstack-barbican | 23:19 | |
*** david-lyle has quit IRC | 23:21 | |
*** mixos has joined #openstack-barbican | 23:29 | |
*** david-ly_ is now known as david-lyle | 23:31 | |
*** kebray has quit IRC | 23:42 | |
*** stevemar_ has quit IRC | 23:47 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!