| *** berndbausch has joined #heat | 00:14 | |
| berndbausch | Good morning, Heat community. | 00:16 |
|---|---|---|
| berndbausch | I came here for a specific question. I have an autoscaling stack that uses cpu_util to detect application load. In Stein, cpu_util and all other derived Ceilometer metrics have disappeared. What metric would you use for autoscaling? | 00:17 |
| berndbausch | I know that Gnocchi has rate aggregations now, but I don't know how to use them with Aodh/Heat. | 00:18 |
| *** ivve has quit IRC | 00:21 | |
| *** hongbin has joined #heat | 00:34 | |
| *** maddtux has joined #heat | 01:00 | |
| *** ricolin has joined #heat | 01:02 | |
| *** ricolin_ has joined #heat | 01:17 | |
| *** ricolin has quit IRC | 01:20 | |
| *** ricolin_ is now known as ricolin | 02:27 | |
| *** ramishra has joined #heat | 03:27 | |
| *** hongbin has quit IRC | 03:49 | |
| gregwork | is there a way to assign a predictable set of ips to a stack programmatically .. rather than hardcode the ips | 03:52 |
| gregwork | eg if i had a list of ips, to start moving down the list | 03:53 |
| *** AJaeger has left #heat | 04:20 | |
| *** skramaja has joined #heat | 04:27 | |
| gregwork | can you initialize %index% as some value | 05:10 |
| gregwork | like 1 | 05:10 |
| *** mlycka has joined #heat | 05:23 | |
| *** jtomasek has joined #heat | 06:01 | |
| ricolin | gregwork, use yaql function we added since newton to add two values https://docs.openstack.org/heat/rocky/template_guide/hot_spec.html#newton | 06:04 |
| *** jtomasek has quit IRC | 06:06 | |
| *** jtomasek has joined #heat | 06:07 | |
| *** ramishra has quit IRC | 06:12 | |
| gregwork | ricolin: i see.. is there a way to extra the security group id from a resource you just created ? .. i think get_attr, but i cant see where the attributes are documented for that | 06:34 |
| ricolin | gregwork, use get_resource: security_group_name to get physical id of a security group | 06:51 |
| mlycka | Hi, I'm back to beg for +w code reviews for my commits, anyone up for it? | 06:59 |
| *** rcernin has quit IRC | 07:03 | |
| *** ramishra has joined #heat | 07:08 | |
| ricolin | mlycka, I trust Ivan's +2 on heat-dashboard, will give few days to see if we can get people from Horizon core team to review them, if not, I will | 07:10 |
| ricolin | mlycka, works for you? | 07:11 |
| mlycka | ricolin: Yep, thank you kindly. | 07:11 |
| * ricolin think we really can use some core member for heat and heat-dashboard:) | 07:15 | |
| berndbausch | Hi everybody, can you help with this question? What metric would you use for autoscaling in Stein, now that cpu_util doesn't exist anymore? | 07:20 |
| openstackgerrit | zhufl proposed openstack/heat master: Fix invalid assert state https://review.opendev.org/668115 | 07:21 |
| berndbausch | Assuming that CPU is the main driver for autoscaling. | 07:22 |
| ricolin | berndbausch, will `compute.node.cpu.percent` works for you? | 07:27 |
| ricolin | #link https://docs.openstack.org/ceilometer/latest/admin/telemetry-measurements.html | 07:27 |
| berndbausch | ricolin, sorry no. This measures the CPU utilization of the hypervisor host, not the instance. But that's exactly what one would need at instance level... | 07:34 |
| *** ivve has joined #heat | 07:34 | |
| berndbausch | The Ceilometer team points to Gnocchi, saying that Gnocchi now has rate aggregation, which allows me to derive the CPU utilization. | 07:35 |
| berndbausch | THis is correct, but | 07:35 |
| berndbausch | Aodh, as far as I can see it, is unable to take advantage of this rate feature. | 07:36 |
| berndbausch | thus, not usable for alarms, thus not usable for scaling. I hope I am wrong though, because it would be a major annoyance. | 07:36 |
| ricolin | berndbausch, just to make, will this works for you or not anymore? https://opendev.org/openstack/heat-templates/src/branch/master/hot/autoscaling.yaml#L138 | 07:44 |
| berndbausch | The problem is line 141, metric:cpu_util. | 07:45 |
| berndbausch | There is no such metric anymore. | 07:45 |
| berndbausch | i have a STein devstack lying around, will try it and see what symptoms I get. | 07:46 |
| ricolin | berndbausch, If you only looking for related functionality and doesn't need to be ceilometer | 07:47 |
| ricolin | you can also check out autoscaling with monasca | 07:48 |
| berndbausch | this is for a training course, and Monasca is too complex :) | 07:48 |
| ricolin | here's example for it https://github.com/sjamgade/monasca-autoscaling | 07:48 |
| berndbausch | The autoscaling.yaml stack requires internet access, which I currently don't have. I will try a simpler one that uses cpu_util | 07:49 |
| *** spsurya has joined #heat | 07:50 | |
| ricolin | berndbausch, you can get signal_url attribute from Heat scale policy and send get request every time you wish to scale, so you can use whatever metering tools you like as long as it can trigger a url request | 08:00 |
| *** jtomasek has quit IRC | 08:02 | |
| berndbausch | The problem is on the alarm side. Prior to Stein, one could set up alarms based on cpu_util. This doesn't work anymore, because Ceilometer removed that metric. Not Heat's fault :) but I though you were aware and had a solution. | 08:04 |
| *** mikecmpbll has joined #heat | 08:05 | |
| berndbausch | The bad thing is that there is no error message anywhere. You can ask Gnocchi for a bogus metric, it will just return an HTTP 200 with an empty data set. | 08:06 |
| berndbausch | Anyway, I will switch to the telemetry IRC. Thanks much for your patience, ricolin. | 08:06 |
| *** berndbausch has left #heat | 08:07 | |
| ricolin | berndbausch, no solution from me since I just aware of and still looking, will ask others in auto-scaling sig if they got any solution for that before I done with my survey. | 08:10 |
| ricolin | probably I should file a story in https://storyboard.openstack.org/#!/project/openstack/auto-scaling-sig | 08:11 |
| *** e0ne has joined #heat | 08:18 | |
| *** mikecmpbll has quit IRC | 08:35 | |
| *** mikecmpbll has joined #heat | 08:36 | |
| *** jtomasek has joined #heat | 08:52 | |
| *** gfidente has joined #heat | 09:36 | |
| *** ricolin_ has joined #heat | 10:01 | |
| *** gmoro has joined #heat | 10:01 | |
| *** zaneb has quit IRC | 10:03 | |
| *** zbitter has joined #heat | 10:03 | |
| *** ricolin has quit IRC | 10:03 | |
| *** maddtux has quit IRC | 10:29 | |
| *** ekultails has joined #heat | 11:59 | |
| *** mhen has joined #heat | 12:11 | |
| mhen | hello there :) | 12:19 |
| mhen | in the Heat template designer in Horizon I am able to create resources (e.g. servers, volumes etc.) and configure them to depend on each other | 12:19 |
| mhen | however, I can't seem to find a way to actually use one resource in the template as the base for another, e.g. create a Cinder volume and then boot a Nova server from that | 12:19 |
| mhen | the selection box for the source volume of the Nova server doesn't list the new volume specified in the very same template even though the latter is marked as dependency for the server | 12:20 |
| mhen | I know it is possible in the template language itself - is this a known limitation of the designer UI or am I missing something? | 12:20 |
| *** ricolin_ is now known as ricolin | 12:48 | |
| ricolin | mhen, I don't have horizon environment with me now, but I think it's possible that the UI need some improvement. We do need more hand on improving designer UI | 12:58 |
| mhen | ricolin, I see. Thanks for the clarification. | 13:02 |
| mhen | also on queens I experience cases where the designer UI will generate invalid templates, i.e. containing attributes not accepted by / known to Heat | 13:06 |
| mhen | e.g. an attribute called "boot_source" for Cinder volumes | 13:06 |
| mhen | using queens/stable branch for both Heat and the Heat dashboard extension | 13:07 |
| mhen | is this known? | 13:09 |
| ricolin | mhen, I think you're the first to report this, and I think it's a bug;/ | 13:10 |
| *** jcoufal has joined #heat | 13:15 | |
| zbitter | I haven't used the horizon GUI, but I'm pretty sure that you should be able to connect a cinder volume to a nova server in the gui. the second thing sounds like a bug | 13:22 |
| *** zbitter is now known as zaneb | 13:22 | |
| mhen | it *is* possible to connect a cinder volume to a nova server, however that is limited to volumes already existing before the template | 13:26 |
| mhen | volumes that are part of the template itself are not listed as an option to be selected, only existing ones are | 13:27 |
| mhen | that seems to be the case for both a) booting from volume as well as b) adding volumes attachment to servers | 13:27 |
| mhen | sadly, I only have a queens setup available atm, so I can't speak for recent releases | 13:28 |
| zaneb | I meant a volume in the template | 13:37 |
| zaneb | I feel confident that there is a way to reference other resources in the template from the template designer, because that is literally the entire point of it | 13:38 |
| zaneb | the mechanics of it I have no idea though because I've never used it | 13:38 |
| *** cliffparsons has joined #heat | 14:04 | |
| *** jdwidari has joined #heat | 14:28 | |
| *** mlycka has quit IRC | 14:30 | |
| *** altlogbot_1 has quit IRC | 14:46 | |
| *** altlogbot_0 has joined #heat | 14:46 | |
| jdwidari | Hi | 14:50 |
| jdwidari | What does the global_index policy allow access to? | 14:51 |
| ricolin | jdwidari, global index allow access to a API to list across all projects if this is what you asking | 15:04 |
| jdwidari | Thanks ricolin | 15:10 |
| gregwork | ricolin: thanks for the tip! | 15:32 |
| jdwidari | global_index is required for 'global_tenant' right? https://docs.openstack.org/api-ref/orchestration/v1/?expanded=list-stacks-detail#id17 | 15:34 |
| jdwidari | What we want to do is to list heat stack for a specific tenant with a token scoped to admin | 15:35 |
| jdwidari | it doesn't seem that either python-openstackclient or python-heatstackclient have an option for that so we're looking at the API | 15:38 |
| jdwidari | openstack stack list does shows stacks from all projects, but doesn't provide an option to list for a specific project | 15:41 |
| jdwidari | openstack stack list --long gives us the global_index error - we have it set to the default to deny for everybody | 15:42 |
| jdwidari | I guess my question is, would global_index allow us to list heat stacks and their owners? | 16:03 |
| ricolin | gregwork, NP:) | 16:05 |
| *** skramaja has quit IRC | 16:14 | |
| *** ricolin_ has joined #heat | 16:23 | |
| *** ivve has quit IRC | 16:23 | |
| *** ricolin has quit IRC | 16:26 | |
| openstackgerrit | Merged openstack/heat master: Change HOST_IP to SERVICE_HOST https://review.opendev.org/673777 | 16:29 |
| *** ricolin_ is now known as ricolin | 16:29 | |
| ricolin | jdwidari, I believe this is what you're ask for `openstack stack list --property tenant=$Specific_project_ID` | 16:30 |
| jdwidari | Oh, let me try that out, thank you | 16:31 |
| ricolin | jdwidari, NP:) | 16:31 |
| *** gfidente has quit IRC | 16:38 | |
| *** ricolin has quit IRC | 16:40 | |
| *** mikecmpbll has quit IRC | 16:40 | |
| *** ramishra has quit IRC | 16:57 | |
| *** e0ne has quit IRC | 17:06 | |
| *** altlogbot_0 has quit IRC | 17:17 | |
| *** altlogbot_2 has joined #heat | 17:22 | |
| *** altlogbot_2 has quit IRC | 17:31 | |
| *** altlogbot_2 has joined #heat | 17:34 | |
| *** spsurya has quit IRC | 18:22 | |
| *** k_mouza has joined #heat | 18:37 | |
| *** e0ne has joined #heat | 18:49 | |
| *** k_mouza has quit IRC | 19:02 | |
| *** jcoufal_ has joined #heat | 19:15 | |
| *** jcoufal has quit IRC | 19:16 | |
| *** jcoufal_ has quit IRC | 19:19 | |
| *** dopplergange has quit IRC | 19:36 | |
| *** e0ne has quit IRC | 19:46 | |
| *** ivve has joined #heat | 20:15 | |
| *** jcoufal has joined #heat | 21:08 | |
| *** dopplergange has joined #heat | 21:12 | |
| *** jcoufal has quit IRC | 21:25 | |
| *** jtomasek has quit IRC | 21:26 | |
| *** mchlumsky has quit IRC | 21:44 | |
| *** stevebaker has joined #heat | 21:46 | |
| *** rcernin has joined #heat | 22:15 | |
| *** cliffparsons has quit IRC | 22:29 | |
| *** ekultails has quit IRC | 22:38 | |
| *** ivve has quit IRC | 22:45 | |
| *** hoonetorg has quit IRC | 23:26 | |
| *** hoonetorg has joined #heat | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!