| *** julim has joined #openstack-helm | 00:31 | |
| *** julim has quit IRC | 00:50 | |
| *** yamamoto has quit IRC | 00:53 | |
| *** yamamoto has joined #openstack-helm | 00:53 | |
| anticw | mlivadariu: works well to get started | 00:59 |
|---|---|---|
| *** cNilesh has joined #openstack-helm | 02:03 | |
| *** unicell has quit IRC | 02:07 | |
| *** pcarver_ has joined #openstack-helm | 02:10 | |
| *** pcarver has quit IRC | 02:12 | |
| *** ianw_pto has quit IRC | 02:12 | |
| *** ianw has joined #openstack-helm | 02:13 | |
| *** gkadam has joined #openstack-helm | 02:26 | |
| *** cfriesen has joined #openstack-helm | 02:31 | |
| *** julim has joined #openstack-helm | 02:33 | |
| *** julim has quit IRC | 04:23 | |
| *** julim has joined #openstack-helm | 04:25 | |
| *** cfriesen has quit IRC | 04:38 | |
| *** mdih has quit IRC | 04:43 | |
| *** gkadam has quit IRC | 04:45 | |
| *** unicell has joined #openstack-helm | 05:06 | |
| *** unicell1 has joined #openstack-helm | 05:13 | |
| *** unicell has quit IRC | 05:13 | |
| *** gkadam has joined #openstack-helm | 05:29 | |
| *** mdih has joined #openstack-helm | 05:45 | |
| *** evin has joined #openstack-helm | 06:10 | |
| *** lunarlamp has quit IRC | 06:44 | |
| *** lunarlamp has joined #openstack-helm | 06:45 | |
| *** girish has joined #openstack-helm | 07:10 | |
| girish | Hi Everyone | 07:10 |
| girish | I tried helm to install openstack onkubernetes | 07:10 |
| girish | I followed the documentation at https://docs.openstack.org/openstack-helm/latest/install/multinode.html | 07:10 |
| girish | But mariadb is not coming up | 07:10 |
| girish | it is indefinitely waiting | 07:10 |
| girish | Can a nybody help me please? | 07:11 |
| girish | ubuntu@worker1:~$ kubectl logs -n openstack mariadb-0 Error from server (BadRequest): container "mariadb" in pod "mariadb-0" is waiting to start: PodInitializing | 07:11 |
| girish | ubuntu@worker1:~$ kubectl logs -n ceph ceph-rgw-85d66f9658-9ds5r container_linux.go:247: starting container process caused "process_linux.go:359: container init caused \"rootfs_linux.go:54: mounting \\\"/var/lib/kubelet/pods/779d4894-4871-11e8-bb36-020d687e4ca0/volume-subpaths/ceph-bin/ceph-rgw/0\\\" to rootfs \\\"/var/lib/docker/aufs/mnt/8814753f0501c502327689cb84187edc1c52965f4ef73acae6166481ef08aa5f\\\" at \\\"/var/lib/dock | 07:12 |
| *** yamamoto has quit IRC | 07:16 | |
| *** yamamoto has joined #openstack-helm | 07:17 | |
| jayahn | portdirect: r u following up with container white paper? | 07:19 |
| *** gkadam has quit IRC | 07:56 | |
| *** yamamoto has quit IRC | 08:11 | |
| *** gkadam has joined #openstack-helm | 08:20 | |
| *** yamamoto has joined #openstack-helm | 08:27 | |
| *** yamamoto has quit IRC | 10:55 | |
| *** lunarlamp has quit IRC | 11:03 | |
| *** lunarlamp has joined #openstack-helm | 11:11 | |
| *** julim has quit IRC | 11:31 | |
| *** yamamoto has joined #openstack-helm | 11:50 | |
| *** yamamoto has quit IRC | 11:50 | |
| *** girish has quit IRC | 12:08 | |
| *** cNilesh has quit IRC | 12:11 | |
| *** yamamoto has joined #openstack-helm | 12:18 | |
| *** mdih has quit IRC | 12:33 | |
| *** cfriesen has joined #openstack-helm | 13:27 | |
| *** yamamoto has quit IRC | 13:56 | |
| *** evin has quit IRC | 14:14 | |
| *** spiette has joined #openstack-helm | 14:56 | |
| *** yamamoto has joined #openstack-helm | 14:57 | |
| *** felipemonteiro_ has joined #openstack-helm | 14:58 | |
| *** felipemonteiro__ has joined #openstack-helm | 14:59 | |
| *** yamamoto has quit IRC | 15:03 | |
| *** felipemonteiro_ has quit IRC | 15:03 | |
| *** jistr|mtgs is now known as jistr | 15:04 | |
| *** spiette has quit IRC | 15:06 | |
| osh-chatbot3 | <alianlianlianlianlian> Hi, I write a email-template.tmpl in openstack-helm-infra/prometheus-alert manager/templates/etc/_email-template.tmpl.tpl, and load it via config map to mount in alertmanager template dir. But when I install in debug mode, the alert-template.tmpl has no context. why? helm cannot load tpl with special char? | 15:09 |
| osh-chatbot3 | <alianlianlianlianlian> My template file as below: # cat prometheus-alertmanager/templates/etc/_alert-templates.tmpl.tpl {{ define "__ecms_email_text" }} <table> {{ range .Alerts }} <tr><td>{{ range .Labels.SortedPairs }}{{ if eq .Name "alertname" }}{{ .Value }}{{ end }}{{ end }}</td></tr> <tr><td>{{ range .Labels.SortedPairs }}{{ if eq .Name "severity" }}{{ .Value }}{{ end }}{{ end }}</td></tr> <tr><td>{{ range .Labels.SortedPairs }}{{ if | 15:11 |
| osh-chatbot3 | eq .Name "alertgroup" }}{{ .Value }}{{ end }}{{ end }}</td></tr> <tr><td>{{ .StartsAt }}</td></tr> <tr><td>{{ range .Annotations.SortedPairs }}{{ if eq .Name "description" }}{{ .Value }}{{ end }}{{ end }}</td></tr> {{ end }} </table> {{ end }} {{ define "__ecms_email_subject" }} {{ range .Alerts }} {{ range .Labels.SortedPairs }}{{ if eq .Name "alertname" }}{{ .Value }}{{ end }}{{ end }} {{ end }} {{ end }} {{ define "email.ecms.subject" | 15:11 |
| osh-chatbot3 | }}{{ template "__ecms_email_subject" . }}{{ end }} {{ define "email.ecms.html" }}{{ template "__ecms_email_text" . }}{{ end }} | 15:11 |
| osh-chatbot3 | <alianlianlianlianlian> Thank you in advance | 15:12 |
| *** julim has joined #openstack-helm | 15:20 | |
| openstackgerrit | Steve Wilkerson proposed openstack/openstack-helm-infra master: WIP: Add ldap support to grafana, update version https://review.openstack.org/563270 | 15:24 |
| openstackgerrit | Tin Lam proposed openstack/openstack-helm master: Add LDAP pool options https://review.openstack.org/560151 | 15:38 |
| *** Guest62064 is now known as lamt | 15:48 | |
| *** spiette has joined #openstack-helm | 15:50 | |
| *** gkadam has quit IRC | 15:53 | |
| *** yamamoto has joined #openstack-helm | 15:59 | |
| *** yamamoto has quit IRC | 16:04 | |
| *** julim has quit IRC | 16:21 | |
| openstackgerrit | Tin Lam proposed openstack/openstack-helm master: Sync mongodb chart https://review.openstack.org/564548 | 16:22 |
| *** julim has joined #openstack-helm | 16:23 | |
| *** felipemonteiro__ has quit IRC | 16:45 | |
| *** felipemonteiro has joined #openstack-helm | 16:45 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Libvirt: add dynamic deps on neutron agents as option. https://review.openstack.org/564561 | 16:54 |
| cfriesen | what is the proposed mechanism for openstack-helm to track resources for use with instances using "dedicated" CPUs? Any host CPUs allocated to such instances must not be used by any k8s pods running on that host. | 16:57 |
| *** yamamoto has joined #openstack-helm | 17:01 | |
| anticw | *not* or *avoided* | 17:01 |
| anticw | i think cpu affinity makes sense, but avoiding things completely causes issues | 17:02 |
| *** yamamoto has quit IRC | 17:06 | |
| *** unicell1 has quit IRC | 17:14 | |
| openstackgerrit | Renis Makadia proposed openstack/openstack-helm master: ceph - split chart into mon, osd and client https://review.openstack.org/559199 | 17:33 |
| *** julim has quit IRC | 17:40 | |
| *** julim has joined #openstack-helm | 17:41 | |
| *** felipemonteiro has quit IRC | 17:51 | |
| *** felipemonteiro has joined #openstack-helm | 17:51 | |
| *** unicell has joined #openstack-helm | 17:52 | |
| *** yamamoto has joined #openstack-helm | 18:02 | |
| *** yamamoto has quit IRC | 18:07 | |
| *** felipemonteiro_ has joined #openstack-helm | 18:17 | |
| *** felipemonteiro has quit IRC | 18:21 | |
| *** evin has joined #openstack-helm | 18:21 | |
| *** julim has quit IRC | 18:26 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Deployments: Move all deployment to be HA by default https://review.openstack.org/564584 | 18:52 |
| openstackgerrit | Chris Wedgwood proposed openstack/openstack-helm master: [RFC] neutron: default to OVSHybridIptablesFirewallDriver firewall driver https://review.openstack.org/564588 | 19:03 |
| anticw | SamYaple (or indeed anyone else with experience here) what do you think about ^ ? | 19:03 |
| *** yamamoto has joined #openstack-helm | 19:04 | |
| anticw | perhaps the commit message should reference https://docs.openstack.org/neutron/pike/admin/config-ovsfwdriver.html | 19:05 |
| *** yamamoto has quit IRC | 19:09 | |
| openstackgerrit | Renis Makadia proposed openstack/openstack-helm master: ceph - split chart into mon, osd and client https://review.openstack.org/559199 | 19:38 |
| openstackgerrit | Renis Makadia proposed openstack/openstack-helm master: ceph - split chart into mon, osd and client https://review.openstack.org/559199 | 19:42 |
| *** felipemonteiro__ has joined #openstack-helm | 19:46 | |
| *** felipemonteiro_ has quit IRC | 19:46 | |
| cfriesen | anticw: sorry, just got back from an all-afternoon meeting. I'm trying to wrap my head around how you'd run a compute node with containerized nova-compute that would be able to run instances with dedicated cpus. | 19:52 |
| cfriesen | anticw: in the non-containerized world this is straightforward, you use isolcpus to isolate the host processes on some "platform" CPUs and then tell nova it's allowed to put guests on the remaining ones. | 19:52 |
| cfriesen | anticw: I can't figure out how this would work with kubernetes, since kubectl seems to assume it owns the whole node | 19:53 |
| *** mdih has joined #openstack-helm | 20:01 | |
| *** yamamoto has joined #openstack-helm | 20:05 | |
| anticw | cfriesen: dedicated means what to you in this case? | 20:08 |
| *** yamamoto has quit IRC | 20:11 | |
| cfriesen | anticw: "hw:cpu_policy=dedicated" in the instance flavor | 20:18 |
| cfriesen | anticw: the intent is to map a guest vCPU 1:1 with a host CPU | 20:18 |
| cfriesen | anticw: this will cause qemu to affine the linux thread corresponding to the guest vCPU to run exclusively on that host CPU, and we want to prevent anything else from running on that host CPU | 20:20 |
| anticw | "we want to prevent anything else from running on that host CPU" i think is a bad idea | 20:33 |
| anticw | if you have some work on that physical machine that needs to be run, let it run | 20:33 |
| anticw | even if it's k8s, etc | 20:33 |
| anticw | cpu afinity can steer things away from that sure, but you don't want to prevent it | 20:33 |
| anticw | or else you get resource stavation or locks held for too long (things break) | 20:33 |
| anticw | so use cpu afinity to steer k8s pods away from those cpus, but don't prevent it | 20:34 |
| cfriesen | anticw: we have this working now with a non-containerized environment. my question is how to ensure that no k8s pods will ever run on those host CPUs (while also having k8s report the correct amount of resources for that node). So if I have 8 cpus total and i want to reserve 6 for instances, then k8s should only "see" 2 CPUs for running pods. | 20:41 |
| anticw | cfriesen: and i'm saying that's a bad idea | 20:41 |
| anticw | you don't want to do that | 20:41 |
| anticw | if k8s needs to do something, and you prevent it running ... badness will happen | 20:41 |
| anticw | so don't prevent it, just let cpu afinity have k8s avoid those cpus | 20:42 |
| cfriesen | anticw: when you say | 20:42 |
| cfriesen | "let cpu affinity have k8s avoid those cpus", are you talking about just using taskset or something? | 20:43 |
| anticw | whatever mechanism works best, i haven't looked at what makes sense in this case | 20:43 |
| cfriesen | anticw: if I use taskset, won | 20:43 |
| anticw | allow k8s to do stuff if it has to, have it avoid the cpus you dedicate for VMs | 20:43 |
| cfriesen | won't k8s still report to the master node that it has 8 cpus when really it can only run on 2? | 20:43 |
| *** evin has quit IRC | 20:43 | |
| anticw | probably | 20:44 |
| anticw | from a resource account PoV perhaps you can have the nova ds on each host 'claim' 6 CPUs? | 20:45 |
| cfriesen | that has potential, I think | 20:46 |
| anticw | your concern is that in a mixed env you'll have expensive pods running taking cycles away from kvm processes? | 20:48 |
| anticw | if so, can you nice things so that kvm gets priority? | 20:49 |
| cfriesen | anticw: in the context of things like software routers we really cannot afford *anything* else running on those host CPUs. (This is a typical NFV usecase, and the cloud operator will generally do significant work to tune the host to ensure that OS-level work gets done on other host CPUs.) | 20:51 |
| *** eeiden has joined #openstack-helm | 20:51 | |
| cfriesen | anticw: in the non-containerized environment, the rest of the system is affined to a small number of pCPUs, and only qemu threads are allowed to run on the "dedicated" pCPUs | 20:52 |
| *** jdandrea has quit IRC | 20:52 | |
| *** yamamoto has joined #openstack-helm | 21:07 | |
| *** yamamoto has quit IRC | 21:13 | |
| *** cfriesen has quit IRC | 21:22 | |
| *** eeiden has quit IRC | 21:33 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Gate: add basic cinder tests to gate https://review.openstack.org/564632 | 21:44 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Gate: add basic cinder tests to gate https://review.openstack.org/564632 | 21:44 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: Deployments: set max_unavailible to 0 https://review.openstack.org/564633 | 21:58 |
| *** yamamoto has joined #openstack-helm | 22:09 | |
| *** yamamoto has quit IRC | 22:14 | |
| *** felipemonteiro__ has quit IRC | 22:28 | |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: WIP: Bootstrap: heat https://review.openstack.org/555932 | 22:39 |
| openstackgerrit | Pete Birley proposed openstack/openstack-helm master: WIP: Bootstrap: heat https://review.openstack.org/555932 | 23:01 |
| *** yamamoto has joined #openstack-helm | 23:11 | |
| *** yamamoto has quit IRC | 23:16 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!