| *** yangyapeng has quit IRC | 00:01 | |
| *** yangyapeng has joined #openstack-helm | 01:13 | |
| *** d|k` has joined #openstack-helm | 01:22 | |
| *** yangyapeng has quit IRC | 01:25 | |
| osh-chatbot1 | <vietwow> Hi all, Can I customize the components of OpenStack-Helm when installing? I mean, I want to use OpenStack-Helm to install all openstack componenet except ceph and mysql, because these are stateful services, so I want to keep it out k8s to safety | 01:36 |
|---|---|---|
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: DNM: Armada based deploy of OSH on Promenade based K8s Cluster https://review.openstack.org/549665 | 01:45 |
| *** mdih has quit IRC | 02:14 | |
| cmm_ | hrm, that's irritating. In nova, pci_alias isn't a list of aliases. I'm stuck with one alias. :-( | 02:49 |
| portdirect | cmm_: its a multistring i think? | 02:51 |
| cmm_ | portdirect, one would think so, but the VM won't schedule when I do it. | 02:51 |
| cmm_ | Unexpected API Error. Please report this at http://bugs.launchpad.net/nova/ and attach the Nova API log if possible. <class 'nova.exception.PciInvalidAlias'> (HTTP 500) (Request-ID: req-2bdb696c-bb3b-4062-81b8-b6ce72613f21) | 02:51 |
| cmm_ | 2018-03-05 02:06:36.984 1 ERROR nova.api.openstack.extensions PciInvalidAlias: Invalid PCI alias definition: [{u'vendor_id': u'8086', u'product_id': u'10e8', u'device_type': u'type-PF', u'name': u'1G'}] is not of type 'object' | 02:52 |
| portdirect | cmm_: what are you feeding into you nova.conf? | 02:53 |
| cmm_ | let me look | 02:54 |
| portdirect | and whats the rendered doc looking like? | 02:54 |
| cmm_ | portdirect, rendered, it looks like this, under [DEFAULT] section | 02:56 |
| cmm_ | pci_alias = {"vendor_id": "8086", "product_id": "10e8", "device_type": "type-PF", "name": "1G"} | 02:56 |
| cmm_ | that's when it *works* | 02:57 |
| cmm_ | when it doesn't work | 02:57 |
| cmm_ | pci_alias = [{"vendor_id": "8086", "product_id": "10e8", "device_type": "type-PF", "name": "1G"}] | 02:57 |
| portdirect | and how are you feeding in the values.yaml ? | 02:58 |
| cmm_ | so, I'm not sure how to specify multiple pci_alias lines in the yaml when passing it into helm. | 02:58 |
| portdirect | unless there has been a regression this should help: https://docs.openstack.org/openstack-helm/latest/devref/oslo-config.html | 02:58 |
| cmm_ | pci_alias: "{{ nova_pci_alias | to_json }}" | 02:58 |
| cmm_ | it's a long story, but I've written ansible playbooks to interface with helm | 02:59 |
| portdirect | nice - was what we had in -infra any help? | 02:59 |
| cmm_ | Unfortunately, I didn't look at -infra when I started. And I'm using Kubespray instead of -infra | 02:59 |
| cmm_ | so I have a bit of a frankenstein monster to set all this up | 03:00 |
| portdirect | the playbooks dont need to use the kubeadm setup | 03:00 |
| portdirect | you just decalre the charts you want to deploy: https://github.com/openstack/openstack-helm-infra/blob/master/tools/gate/chart-deploys/default.yaml | 03:00 |
| portdirect | then run `make dev-deploy charts` | 03:01 |
| portdirect | https://github.com/openstack/openstack-helm-infra/blob/master/tools/gate/devel/start.sh#L70 | 03:01 |
| cmm_ | ic | 03:01 |
| cmm_ | yeah, I was still a noob (still am) to helm, so I wasn't too comfortable with it | 03:02 |
| cmm_ | regarding this link, https://docs.openstack.org/openstack-helm/latest/devref/oslo-config.html how does it help me? | 03:02 |
| cmm_ | I understand how conf: and DEFAULT: work, but I'm not sure how to specify two of the same thing (pci_alias) | 03:03 |
| cmm_ | since YAML would probably only take one of them | 03:03 |
| portdirect | check out the `oslo_messaging_notifications` in the link i gave you | 03:04 |
| portdirect | https://docs.openstack.org/openstack-helm/latest/devref/oslo-config.html | 03:04 |
| portdirect | fr an example of how to do this | 03:04 |
| portdirect | *for | 03:04 |
| cmm_ | oh! neat! | 03:04 |
| cmm_ | got it | 03:05 |
| cmm_ | :-) | 03:05 |
| cmm_ | Man you've thought of everything | 03:05 |
| portdirect | we had to hack round it | 03:05 |
| * portdirect shakes fist at silly multistrings! | 03:05 | |
| cmm_ | lol | 03:05 |
| *** mdih has joined #openstack-helm | 03:06 | |
| cmm_ | can't wait for SR-IOV... I've said this already. :-) In the meantime, trying PCI forwarding. I need direct access to a network interface because of the security testing we do. 802.1 MAC addresses don't flow through OVS by default. | 03:07 |
| cmm_ | portdirect, indeed. type: multistring works. thank you | 03:21 |
| cmm_ | pci_alias = {"vendor_id": "8086", "product_id": "10e8", "device_type": "type-PF", "name": "1G"} | 03:21 |
| cmm_ | pci_alias = {"vendor_id": "8086", "product_id": "1583", "device_type": "type-PF", "name": "40G"} | 03:21 |
| cmm_ | w00t first pci forwarded instance up and running. :-) I'm off to bed | 04:26 |
| *** cmm_ has quit IRC | 04:27 | |
| *** yangyapeng has joined #openstack-helm | 07:02 | |
| *** unicell has joined #openstack-helm | 07:16 | |
| *** unicell has quit IRC | 08:33 | |
| *** yangyape_ has joined #openstack-helm | 09:18 | |
| *** yangyapeng has quit IRC | 09:18 | |
| *** MarkBaker has joined #openstack-helm | 09:22 | |
| *** MarkBaker has quit IRC | 12:20 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: WIP: Neutron SR-IOV support https://review.openstack.org/547890 | 12:29 |
| *** julim has quit IRC | 12:31 | |
| *** MarkBaker has joined #openstack-helm | 12:32 | |
| *** yangyape_ has quit IRC | 13:19 | |
| *** MarkBaker has quit IRC | 13:34 | |
| *** ianychoi has joined #openstack-helm | 13:41 | |
| *** MarkBaker has joined #openstack-helm | 13:47 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Keystone Authtoken Cache: allow universal secret key to be set https://review.openstack.org/549607 | 13:49 |
| *** jistr is now known as jistr-mtg | 13:49 | |
| *** julim has joined #openstack-helm | 14:03 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: RabbitMQ: persist stateful data to volume https://review.openstack.org/549772 | 14:23 |
| *** mdih has quit IRC | 14:43 | |
| *** mdih has joined #openstack-helm | 14:45 | |
| *** mdih has quit IRC | 14:46 | |
| openstackgerrit | Scott Hussey proposed openstack/openstack-helm master: Ingress - Add configmap hashes to deployments https://review.openstack.org/549778 | 14:50 |
| *** cmm_ has joined #openstack-helm | 14:53 | |
| *** julim_ has joined #openstack-helm | 14:54 | |
| *** haobing has joined #openstack-helm | 14:56 | |
| *** julim has quit IRC | 14:57 | |
| d|k` | portdirect: about addressing pci et al. disks for ceph: would it be reasonable to use, say, IDs as under /dev/disk/by-id? | 14:58 |
| d|k` | portdirect i ask because it seems like all the PCI names i can find (ie, under /dev/disk/by-path/pci-*) are more-or-less just the bus-specific (scsi, frex) names with the pci location of the controller bunged on the front | 14:59 |
| d|k` | portdirect: ... which is really just kinda a more-verbose way of specifying by scsi bus path. | 14:59 |
| d|k` | portdirect: i'm thinking here of, say, a 'bus-id' type where the location is the id. | 15:00 |
| d|k` | errr, block-id, but you get the id. | 15:00 |
| d|k` | errr, block-id, but you get the idea. halp, too early, caffeine not yet working. | 15:00 |
| *** d|k` is now known as d|k | 15:01 | |
| *** d|k is now known as Guest11793 | 15:01 | |
| *** Guest11793 is now known as d|k` | 15:02 | |
| *** d|k` has left #openstack-helm | 15:05 | |
| *** d|k` has joined #openstack-helm | 15:05 | |
| *** d|k` is now known as d|k | 15:07 | |
| *** d|k has quit IRC | 15:08 | |
| *** d|k has joined #openstack-helm | 15:09 | |
| *** haobing has quit IRC | 15:13 | |
| *** jistr-mtg is now known as jistr | 15:16 | |
| *** openstackgerrit has quit IRC | 15:34 | |
| *** yangyapeng has joined #openstack-helm | 15:40 | |
| *** openstackgerrit has joined #openstack-helm | 15:40 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: RabbitMQ: update to version 3.7.3 https://review.openstack.org/549792 | 15:40 |
| *** yangyapeng has quit IRC | 15:44 | |
| cmm_ | d|k nobody asked me, but I've found that world-wide-names in /dev/disk/by-id/ are great for identifying storage devices. | 15:54 |
| cmm_ | and it'll be pretty static | 15:54 |
| openstackgerrit | Hemanth Nakkina proposed openstack/openstack-helm-infra master: Remove Api version requirement for CronJob Curator https://review.openstack.org/549795 | 16:00 |
| d|k | cmm_: yah, i think it'd be a super useful addition to our plans (here: https://review.openstack.org/545298 ) to do scsi/iscsi addressing | 16:01 |
| d|k | and yeah, the wwn IDs were exactly what I had in mind. | 16:01 |
| d|k | they also have a standard way of naming partitions, which [i]scsi paths don't. | 16:02 |
| cmm_ | I'm not a storage expert, but even LUNs on a RAID get a WWN. They're like a MAC address. They might be random, but the controller makes them | 16:02 |
| cmm_ | iSCSI is something else though. :-\ | 16:02 |
| d|k | i _think_ even iscsi devices might get a WWN once attached, but i'm not positive about that. | 16:03 |
| d|k | ... and at least at present we're just considering already-attached devices. | 16:03 |
| cmm_ | partitions on WWNs also get a -partX suffix | 16:04 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Elasticsearch: use endpoints section and lookups to set port https://review.openstack.org/549799 | 16:04 |
| d|k | cmm_: yeah, exactly. annoyingly, there seems to be no such suffix for, say, scsi paths. | 16:06 |
| cmm_ | oh :-( | 16:06 |
| d|k | ... but if you look at the WWN view under by-path, they have them, so that's a good feature. | 16:06 |
| cmm_ | ah. I wonder if udev/systemd needs an enhancement for that | 16:07 |
| d|k | for scsi, i mean if you look under, say, /sys/bus/scsi you don't find partitions. | 16:07 |
| cmm_ | or a rule file created...something like that | 16:07 |
| d|k | so if you have a scsi location like 'scsi@0:7.0' it's a _little_ indirect to get to partitions on that device | 16:08 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Fluent-logging: use endpoints section and lookups to set port https://review.openstack.org/549801 | 16:08 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Prometheus: use endpoints section and lookups to set port https://review.openstack.org/549802 | 16:11 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Grafana: use endpoints section and lookups to set port https://review.openstack.org/549803 | 16:17 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Kibana: use endpoints section and lookups to set port https://review.openstack.org/549806 | 16:19 |
| *** MarkBaker has quit IRC | 16:29 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: WIP: support local image registry https://review.openstack.org/545642 | 16:32 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Alertmanager: use endpoints section and lookups to set port https://review.openstack.org/549811 | 16:33 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: kube-state-metrics: use endpoints section and lookups to set port https://review.openstack.org/549815 | 16:39 |
| *** MarkBaker has joined #openstack-helm | 16:43 | |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: node-exporter: use endpoints section and lookups to set port https://review.openstack.org/549818 | 16:44 |
| openstackgerrit | Merged openstack/openstack-helm master: Horizon: use endpoints section and lookups to set port https://review.openstack.org/549647 | 16:45 |
| *** mdih has joined #openstack-helm | 16:46 | |
| *** jdandrea has quit IRC | 16:47 | |
| *** jdandrea has joined #openstack-helm | 16:48 | |
| coboluxx | alanmeadows: i've updated 548724 with support for referencing an image for a completly custom horizon theme. is this what you had in mind? | 16:50 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Armada: Update manifest to use common memcached service https://review.openstack.org/549612 | 16:56 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Armada: Update manifest to use common memcached service https://review.openstack.org/549612 | 17:01 |
| *** cmm_ has quit IRC | 17:06 | |
| *** cmm_ has joined #openstack-helm | 17:13 | |
| openstackgerrit | Jawon Choo proposed openstack/openstack-helm master: Cinder: add pass message for enabling application https://review.openstack.org/549824 | 17:15 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Fluent-logging: use endpoints section and lookups to set port https://review.openstack.org/549801 | 17:20 |
| openstackgerrit | Merged openstack/openstack-helm master: Congress: use endpoints section and lookups to set port https://review.openstack.org/549492 | 17:20 |
| openstackgerrit | Merged openstack/openstack-helm master: Mistral: use endpoints section and lookups to set port https://review.openstack.org/549491 | 17:24 |
| *** yangyapeng has joined #openstack-helm | 17:39 | |
| *** julim has joined #openstack-helm | 17:41 | |
| *** julim has quit IRC | 17:41 | |
| *** julim has joined #openstack-helm | 17:42 | |
| *** julim_ has quit IRC | 17:43 | |
| *** yangyapeng has quit IRC | 17:44 | |
| openstackgerrit | Merged openstack/openstack-helm master: Nova: use endpoints section and lookups to set port https://review.openstack.org/549498 | 17:54 |
| openstackgerrit | Merged openstack/openstack-helm master: Barbican: use endpoints section and lookups to set port https://review.openstack.org/549478 | 17:54 |
| openstackgerrit | Merged openstack/openstack-helm master: Remove some useless indents https://review.openstack.org/538123 | 17:54 |
| openstackgerrit | Hari proposed openstack/openstack-helm master: Implementes: ranger-agent helm charts https://review.openstack.org/538341 | 18:09 |
| *** felipemonteiro has joined #openstack-helm | 18:09 | |
| *** yangyapeng has joined #openstack-helm | 18:29 | |
| *** yangyapeng has quit IRC | 18:34 | |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Grafana: use endpoints section and lookups to set port https://review.openstack.org/549803 | 18:35 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Alertmanager: use endpoints section and lookups to set port https://review.openstack.org/549811 | 18:36 |
| *** unicell has joined #openstack-helm | 18:56 | |
| *** felipemonteiro has quit IRC | 19:08 | |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Grafana: use endpoints section and lookups to set port https://review.openstack.org/549803 | 19:16 |
| *** mdih has quit IRC | 19:16 | |
| *** mdih has joined #openstack-helm | 19:17 | |
| *** unicell1 has joined #openstack-helm | 19:24 | |
| *** unicell has quit IRC | 19:24 | |
| openstackgerrit | Merged openstack/openstack-helm master: Keystone Authtoken Cache: allow universal secret key to be set https://review.openstack.org/549607 | 19:26 |
| openstackgerrit | Merged openstack/openstack-helm master: RabbitMQ: persist stateful data to volume https://review.openstack.org/549772 | 19:37 |
| openstackgerrit | Hari proposed openstack/openstack-helm master: Implementes: ranger-agent helm charts https://review.openstack.org/538341 | 20:14 |
| *** yangyapeng has joined #openstack-helm | 20:48 | |
| *** yangyapeng has quit IRC | 20:53 | |
| cmm_ | Hi all, out of curiosity, is moving to Ocata, Pike or even Queens considered really hard at this time? What needs to happen to enable images other than Newton, aside from updating the container images? | 21:06 |
| cmm_ | when I say update, I don't mean update an existing cluster... | 21:07 |
| *** felipemonteiro has joined #openstack-helm | 21:20 | |
| lamt | not sure if I will say hard. Barring any backward incompatible changes between OpenStack versions, updating the image and necessary configuration changes *should* be all you need to newly install Pike or Queen. However, it is not tested and OSH doesn't officially support Pike (yet), so ymmv. If you want to try it and post your findings, that'd be great. | 21:30 |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: Alertmanager: use endpoints section and lookups to set port https://review.openstack.org/549811 | 21:32 |
| *** julim has quit IRC | 21:37 | |
| *** felipemonteiro has quit IRC | 22:01 | |
| openstackgerrit | Merged openstack/openstack-helm master: Armada: Update manifest to use common memcached service https://review.openstack.org/549612 | 22:10 |
| *** gmmaha has quit IRC | 22:11 | |
| *** gmmaharaj has joined #openstack-helm | 22:11 | |
| *** gmmaharaj is now known as gmmaha | 22:11 | |
| cmm_ | lamt, cool, just wondering | 22:14 |
| cmm_ | did we strive to keep compatibility with Kolla, or was that a non-goal? | 22:14 |
| *** julim has joined #openstack-helm | 22:24 | |
| *** cmm_ has quit IRC | 22:36 | |
| *** yamamoto has joined #openstack-helm | 22:44 | |
| *** yamamoto has quit IRC | 22:48 | |
| SamYaple | define compatibility | 23:05 |
| SamYaple | i think kolla images are still first class citizens | 23:05 |
| SamYaple | havent heard plans to change that | 23:06 |
| *** yamamoto has joined #openstack-helm | 23:07 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Magnum: use endpoints section and lookups to set port https://review.openstack.org/549483 | 23:22 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Glance: use endpoints section and lookups to set port https://review.openstack.org/549479 | 23:22 |
| *** yangyapeng has joined #openstack-helm | 23:27 | |
| openstackgerrit | Paul Belanger proposed openstack/openstack-helm-infra master: Update to fedora-27 for testing https://review.openstack.org/549890 | 23:29 |
| *** yangyapeng has quit IRC | 23:32 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!