*** lawrancejing has quit IRC | 00:00 | |
*** pratikmallya has joined #senlin | 00:11 | |
*** pratikmallya has quit IRC | 00:16 | |
*** lawrancejing has joined #senlin | 00:32 | |
*** elynn has joined #senlin | 00:34 | |
*** pratikmallya has joined #senlin | 00:41 | |
*** pratikmallya has quit IRC | 01:12 | |
*** zhenguo has joined #senlin | 01:16 | |
*** Shijia has joined #senlin | 01:23 | |
*** jdandrea has quit IRC | 01:23 | |
*** Yanyanhu has joined #senlin | 01:37 | |
elynn | morning | 01:53 |
---|---|---|
*** lixinhui has joined #senlin | 01:56 | |
*** xuhaiwei has joined #senlin | 02:07 | |
openstackgerrit | Merged openstack/senlin: Zone placement policy https://review.openstack.org/221684 | 02:08 |
openstackgerrit | Merged openstack/senlin: Add project_safe param for cluster/node load https://review.openstack.org/236779 | 02:10 |
openstackgerrit | Merged openstack/senlin: Make profile related DB query project_safe https://review.openstack.org/236785 | 02:10 |
openstackgerrit | Merged openstack/senlin: Updated from global requirements https://review.openstack.org/234681 | 02:32 |
openstackgerrit | xu-haiwei proposed openstack/senlin: Revise cluster_add_nodes action's behavior https://review.openstack.org/235728 | 02:43 |
openstackgerrit | xu-haiwei proposed openstack/senlin: Revise cluster_add_nodes action's behavior https://review.openstack.org/235728 | 02:49 |
*** lawrancejing has quit IRC | 02:53 | |
*** lawrancejing has joined #senlin | 02:54 | |
openstackgerrit | Yanyan Hu proposed openstack/senlin: Make event related DB queries project safe https://review.openstack.org/237378 | 03:05 |
openstackgerrit | Yanyan Hu proposed openstack/senlin: Make trigger related DB queries project safe https://review.openstack.org/237384 | 03:40 |
elynn | Hi a question, can I update min_size or mix_size of a cluster | 03:45 |
Yanyanhu | elynn, yes | 04:01 |
*** lawrancejing has quit IRC | 04:01 | |
Yanyanhu | it can be done through cluster resize API | 04:01 |
*** lawrancejing has joined #senlin | 04:02 | |
Yanyanhu | in cluster update API, only name, metadata, parent, timeout and profile can be updated | 04:03 |
Yanyanhu | hi, elynn, I guess for senlin cluster resource, the updating of different properties may need to be supported using different senlin API interfaces | 04:06 |
*** gongysh has quit IRC | 04:12 | |
elynn | Yanyanhu, I got it. | 04:16 |
*** gongysh has joined #senlin | 04:17 | |
*** xuhaiwei has quit IRC | 04:46 | |
*** mathspanda has joined #senlin | 04:58 | |
*** pratikmallya has joined #senlin | 06:03 | |
*** elynn has quit IRC | 06:06 | |
*** elynn_ has joined #senlin | 06:06 | |
*** Qiming has joined #senlin | 06:06 | |
openstackgerrit | Merged openstack/senlin: Revise cluster_add_nodes action's behavior https://review.openstack.org/235728 | 06:07 |
*** Shijia has quit IRC | 06:09 | |
*** Shijia has joined #senlin | 06:10 | |
*** Liu has joined #senlin | 06:13 | |
*** Shijia_ has joined #senlin | 06:28 | |
*** Shijia has quit IRC | 06:31 | |
*** elynn__ has joined #senlin | 06:57 | |
*** elynn_ has quit IRC | 06:58 | |
*** pratikmallya has quit IRC | 07:34 | |
*** openstackgerrit has quit IRC | 07:46 | |
*** openstackgerrit has joined #senlin | 07:47 | |
*** elynn__ has quit IRC | 08:00 | |
*** shu has quit IRC | 08:13 | |
*** Yanyanhu has quit IRC | 08:15 | |
*** Yanyanhu has joined #senlin | 08:16 | |
*** elynn has joined #senlin | 08:36 | |
elynn | Hi | 08:36 |
elynn | A question about profile update for cluster. | 08:37 |
elynn | Is it able to update profile for a cluster? | 08:37 |
elynn | I got an error saying "Cannot update a cluster to a different profile type, operation aborted" | 08:37 |
Qiming | hi, updating a profile to a different type is not allowed | 08:38 |
Qiming | say you have a cluster of nova servers, and now you want it to contain heat stacks, that is not acceptable | 08:39 |
Qiming | it is perfectly okay to update a cluster to a different profile of the same type | 08:39 |
elynn | That means I can update to a profile with the same type, even though I change the heat template in spec? | 08:40 |
Qiming | yes, whatever template you want to use | 08:41 |
elynn | After update, new nodes will create with new profile while old nodes are not changed? | 08:41 |
Qiming | these are strictly separated actions in senlin | 08:41 |
Qiming | for 'update', it only means you are replacing the profile used by a cluster | 08:42 |
Qiming | for changing the size of a cluster (add/remove nodes), it is not an 'update' | 08:42 |
Qiming | it is called 'resize' | 08:42 |
Qiming | for heat stack update, we have a dependent patch not yet merged: https://review.openstack.org/#/c/225560/ | 08:43 |
elynn | I mean scale-out | 08:43 |
Qiming | yes, scale-out is not called an update either | 08:44 |
Qiming | we deliberately avoided calling all changes an 'update' | 08:44 |
Qiming | that is making things very difficult in Heat | 08:44 |
elynn | If I using cluster-update and change the profile, what will happen? | 08:44 |
Qiming | the cluster is locked | 08:44 |
Qiming | then all nodes in that cluster will be invoked with a do_update() call | 08:45 |
Qiming | the 'do_update()' call will then be passed to the drivers (say nova or heat) to update the profile used | 08:45 |
Qiming | when all those actions complete | 08:45 |
Qiming | cluster update ends and lock is released | 08:45 |
Qiming | during this operation, you won't be able to resize the cluster, for example | 08:46 |
Qiming | it is *locked* for update | 08:46 |
elynn | Ok, I got it. Say if I use os.heat.stack type of profile for cluster, If I do cluster-update to update the a new profile, then all nodes will do stack-update to fit for the new profile. | 08:48 |
elynn | Same for the nova instance, isn't it? | 08:48 |
Qiming | yes | 08:49 |
Qiming | it is designed that way | 08:49 |
Qiming | but I cannot guarantee you the update operation works at the moment, :( | 08:49 |
Qiming | the update behavior is completely up to the backend service (nova, heat ...) | 08:50 |
Qiming | we just call their update APIs to get things done | 08:50 |
elynn | If I update spec for the profile, what will happen to cluster which using this profile? | 08:53 |
Qiming | 'spec' field is not updatable | 08:53 |
Qiming | if you update that field, you will get a new profile created with a new ID | 08:54 |
elynn | old profile will be deleted? | 08:54 |
Qiming | no | 08:54 |
Qiming | when you do PUT /senlin/v1/profiles/<ID> {'spec': 'something new'} | 08:55 |
elynn | Why do we accept it as parameter in profile-update command, since it doesn't behave like usual update. | 08:55 |
Qiming | you are basically saying: hey, create a new profile using this spec, and copy all other properties from the one I specified | 08:56 |
Qiming | it can be forbidden | 08:56 |
Qiming | obviously a trade-off | 08:56 |
elynn | Ok... | 08:56 |
Qiming | it is strange | 08:56 |
*** lawranc__ has joined #senlin | 08:56 | |
Qiming | maybe we should declare it loudly ... profile spec cannot be updated | 08:57 |
elynn | I didn't allow it to update in my spec, so that's ok for me... | 08:57 |
Qiming | cool | 08:57 |
Qiming | and it doesn't hurt to know more about our original thinking | 08:58 |
Qiming | allowing update to profile spec is not a convenience, I see no strong use case of that, only endless problems to solve | 08:58 |
Yanyanhu | I guess maybe we should just forbid profile updating :) | 08:59 |
*** lawrancejing has quit IRC | 08:59 | |
Qiming | right | 08:59 |
Qiming | it is weird that an update request creates a new object with a new ID | 09:00 |
Yanyanhu | yes | 09:00 |
elynn | So the profile name and metadata can not update for now? | 09:00 |
Yanyanhu | and too many potential issues if a profile which has been used to create cluster/node is updated | 09:00 |
Qiming | it is more like: senlin profile-create --spec <file> --copy-properties-from <existing_profile_id> | 09:00 |
elynn | profile-update will always create a new profile? | 09:00 |
Qiming | name and metadata can be updated freely | 09:01 |
Qiming | yes, that sounds a bad design, right? | 09:01 |
Qiming | was trying to be more friendly to users | 09:01 |
Qiming | maybe it was wrong | 09:01 |
Qiming | feel free to propose a patch to kill it | 09:04 |
Qiming | it is okay to add something later, but removing something that is already in would be a huge headache | 09:04 |
elynn | yes, need to figure out how to implement this in another way before we remove it. | 09:09 |
*** lawranc__ has quit IRC | 09:15 | |
*** lawrancejing has joined #senlin | 09:15 | |
*** elynn has quit IRC | 09:30 | |
*** lawrancejing has quit IRC | 09:34 | |
openstackgerrit | Yanyan Hu proposed openstack/senlin: Make trigger related DB queries project safe https://review.openstack.org/237384 | 09:49 |
openstackgerrit | Yanyan Hu proposed openstack/senlin: Add config option for name uniqueness https://review.openstack.org/234051 | 09:49 |
openstackgerrit | Yanyan Hu proposed openstack/senlin: Make webhook related DB query project safe https://review.openstack.org/236841 | 09:49 |
openstackgerrit | Shu Muto proposed openstack/senlin-dashboard: Add STATUS into detail view of node https://review.openstack.org/237495 | 09:53 |
openstackgerrit | venkatamahesh proposed openstack/senlin-dashboard: Change repository from stackforge to openstack https://review.openstack.org/236686 | 09:58 |
*** zhenguo has quit IRC | 10:03 | |
*** lawrancejing has joined #senlin | 10:16 | |
*** Yanyanhu has quit IRC | 10:20 | |
*** Qiming has quit IRC | 10:53 | |
*** gongysh has quit IRC | 10:54 | |
*** Liu has quit IRC | 11:00 | |
*** Qiming has joined #senlin | 11:40 | |
*** lawrancejing has quit IRC | 11:49 | |
*** lixinhui has quit IRC | 11:50 | |
*** pratikmallya has joined #senlin | 12:11 | |
*** Shijia_ has quit IRC | 12:24 | |
*** Shijia has joined #senlin | 12:24 | |
*** gongysh has joined #senlin | 12:33 | |
*** lkarm has joined #senlin | 12:46 | |
*** yanyanhu has joined #senlin | 12:54 | |
*** pratikmallya has quit IRC | 13:00 | |
*** jruano has joined #senlin | 13:01 | |
*** mathspanda has quit IRC | 13:32 | |
*** elynn has joined #senlin | 13:39 | |
*** Shijia has quit IRC | 13:53 | |
*** yanyanhu has quit IRC | 14:00 | |
*** jruano has quit IRC | 14:05 | |
*** elynn has quit IRC | 14:14 | |
*** zhenguo has joined #senlin | 14:16 | |
*** jdandrea has joined #senlin | 14:17 | |
*** pratikmallya has joined #senlin | 14:20 | |
*** pratikma_ has joined #senlin | 14:21 | |
*** pratikmallya has quit IRC | 14:24 | |
*** Shijia has joined #senlin | 15:11 | |
*** gongysh has quit IRC | 15:20 | |
openstackgerrit | yan.haifeng proposed openstack/python-senlinclient: Delete openstack folder to advoid import issue https://review.openstack.org/237637 | 15:21 |
*** Qiming has quit IRC | 15:41 | |
*** Shijia has quit IRC | 16:03 | |
*** pratikma_ has quit IRC | 16:16 | |
*** pratikmallya has joined #senlin | 16:43 | |
*** pratikma_ has joined #senlin | 16:47 | |
*** pratikmallya has quit IRC | 16:50 | |
*** lkarm has quit IRC | 17:45 | |
*** lkarm has joined #senlin | 18:45 | |
*** lkarm has quit IRC | 18:50 | |
*** lkarm has joined #senlin | 19:01 | |
*** lkarm has quit IRC | 19:05 | |
*** lkarm has joined #senlin | 19:09 | |
*** lkarm has quit IRC | 19:45 | |
*** lkarm has joined #senlin | 20:07 | |
*** pratikma_ has quit IRC | 20:20 | |
*** lkarm has quit IRC | 20:33 | |
*** pratikmallya has joined #senlin | 20:45 | |
*** pratikma_ has joined #senlin | 20:46 | |
*** pratikmallya has quit IRC | 20:50 | |
*** lkarm has joined #senlin | 20:59 | |
*** lkarm has quit IRC | 21:04 | |
*** pratikma_ has quit IRC | 22:34 | |
*** jongchoi has joined #senlin | 22:36 | |
*** jongchoi has quit IRC | 22:36 | |
*** pratikmallya has joined #senlin | 23:22 | |
*** pratikmallya has quit IRC | 23:41 | |
*** Qiming has joined #senlin | 23:47 | |
*** Qiming has quit IRC | 23:48 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!