[02:18] #openstack === cesar_ is now known as Guest81200 [02:42] hi [02:43] hello ther [02:43] there [02:43] hello [02:44] hello [02:44] I just have a question [02:44] i need to implement a service [02:44] new service with openstack [02:45] bot i realy don't understand how JUJU can heolpme [02:45] I have been created a BareMetal as a service [02:45] with MaaS [02:46] but I can't understand for example if it's necesary to deploy juju for that [02:47] or if juju can helpme with containers [02:47] ... [02:48] MaaS can give you servers with Ubuntu installed on them [02:48] "bare metal" [02:48] ok [02:48] JuJu is a larger beast, that can install software on top of those servers [02:49] that software is packaged up into "charms" [02:49] disclaimer: I have not used JuJu [02:49] Openstack is software [02:50] ok sooo with maas i can get servers (elastic from the metal) and with juju i can install under the servers all type of software allowed [02:50] the word elastic doesn't mean much to me in that context [02:50] thats right [02:51] if you're looking to deploy openstack, you probably are looking to let your users spin up their own VMs [02:51] not so much to allow them to take control of entire servers (physical hardware) [02:52] what I would expect is that you use something like MaaS to manage your fleet of hardware, then JuJu to manage the deployment of openstack, which will contain a hypervisor and all the control software to allow users to make the hypervisor do stuff for them [02:53] mmm no this service will allow the user to get dedicated server in a BareMetal way with MAAS an JUJU [02:53] that will be the limit of the user [02:55] im traying to implement a service of BareMetal with MAAS for users, but that is gonna be like a VIP service [02:55] dedicated server with that technology [02:57] if you just want to provide bare metal, you can do that without JuJu [02:57] ok [02:57] at my last job I used MaaS without juju [02:58] I was the only user of MaaS though [02:58] I had a lot of trouble getting https to work for the web UI [03:00] was difficult the support while you operate the service? [03:01] I don't understand your question [03:01] I had no engagement with canonical support staff...? [03:03] what i try to say was, if the implementation and operation of the service is complicated [03:03] for you [03:04] in your own experience [03:04] oh ok [03:04] that is a tricky one to answer [03:04] jajaja [03:04] When implementing MaaS I did not have an infrastructure background [03:05] oh [03:05] I am a software developer [03:05] So there was a lot for me to learn, and I made a lot of mistakes [03:05] Hard to tell if it's because the product is complicated, or if I just didn't know enough [03:06] Once i learned it all it made a lot of sense [03:06] ok [03:08] thanks for your help ! [03:09] it was a great to know your opinion [03:12] no worries [03:12] oh [03:12] other thing to note is that this time zone doesn't often have many people in the chat room [03:12] I found it pretty hard to ask anyone questions about how to MaaS properly [03:14] oh [03:15] ok so at what time it's better to be hear? [03:16] *here [03:19] dunno... I think the devs are in UK [06:23] is there a way to logon to a failed commission machine [06:23] i need to debug [06:23] i tried ubuntu/ubuntu [06:28] does this even work [06:28] anyone here? before i search myself to death [06:34] i tried uncommenting /etc/maas/preseeds/enlist_userdata [06:34] and then ubuntu/ubuntu again [06:34] but that doesnt work [07:46] Bug #1613439 changed: [2.0 RC4] failure to access ephemeral image during deployment - iscsistart: initiator reported error (11 - iSCSI PDU timed out) [07:46] Bug #1666719 changed: [2.1.3] A node enlistment fails to contact metadata service === frankban|afk is now known as frankban [08:28] how can i debug a failed commission machine? [08:32] ybaumy, if using a recent version of maas, start by checking for /var/log/maas/rsyslog// [08:36] brendand: too bad there is nothing logged there. it fails before it starts to log with rsyslog [08:36] thats my problem [08:36] i need to logon to the machine somehow [08:37] i setup a test VM with the same networks and everything is working [08:37] so it cannot be networking related as far as i can see [08:38] the strang thing is on default network in maas the commissioning works [08:38] strange [08:49] can i capture screen output in vmware somehow? [08:49] from the console [08:58] it says reached target network so this is working [08:58] but then nothing [08:59] i could also read that no datasource found [09:00] what datasource is meant there? local or internet? [09:00] if local which one [09:00] if internet then the test that was successfull was failed in the end? [09:01] Bug #1673377 opened: [2.2, Hardware Test] Add test for NTP connectivity [09:01] Bug #1673378 opened: [2.2, Hardware Test] Add test for NTP connectivity === frankban is now known as frankban|afk [10:11] what configuration is wrong if a deployed node doesn't have internet access while the maas server does? [10:14] apt works but nothing else. does maas only supply repository access? [10:15] DesktopMan: if apt works you probably have set a apt proxy [10:16] I didn't manually but I guess maas does automatically [10:17] DesktopMan: have you succesfully setup a node in this subnet with internet access before or is it the same as the maas default network? [10:17] default. so I guess I need to do more [10:17] DesktopMan: if so . have you set subnet gateway and dns server? === frankban|afk is now known as frankban [10:18] does the maas server do nat by default? or do I need to set that up [10:19] DesktopMan: i dont think so [10:19] DesktopMan: i setup nat on my asa firewall [10:20] DesktopMan: but now im using internal ntp and apt proxy and everything working [10:31] how does maas compare with openstack ironic? trying to figure out pros/cons [10:32] i use maas to deploy openstack on top of it [10:34] with juju or something else? or did you install openstack manually after deploying from maas [10:34] yes i use juju [10:34] vmware + maas + juju + openstack [10:35] right. do you run the juju controller on the same host as the maas controller? wondering if I should split them [10:36] no. i setup juju controllers on different vm's [10:36] right [10:36] why do you have multiple juju controllers? lots of hardware? [10:36] i have 3 datacenter setup. so i deploy the controllers to each datacenter [10:37] makes sense [10:39] have you looked at Fuel at all? [10:39] wondering how it stacks up to maas+juju [10:39] havent yet [10:40] im still dreaming the dream that i can setup everything with this... but there are always problems [10:41] hehe [10:41] its still a POC [10:42] and i have time [10:42] else i would have switched to suse cloud 7 or redhat already [10:42] yeah we're also just doing exploratory testing right now [10:44] right now im having trouble with commissioning hosts on another subnet. there is no way to troubleshoot this. which is really annoying [10:44] or at least i dont know the way [10:52] morning [11:23] how can I set debugging on for seperate components? [11:24] for e.g: IPMI? where is the IPMI config for or before power on? [11:25] strauss_: there is, but it only tells the machine to PXE boot [11:28] I know. it powers on my machine but stuck in provisioning state [11:29] I want to use IPMI to set next boot PXE in ILO [11:31] so how is the workflow defined? I select action commision node. IPMI powers on node pxe and get bootimage. I more detailed IPMI config [11:34] Bug #1673433 opened: [2.2b3] MAAS reports DHCP is running but it is not [14:13] is there a problem with vm's and multiple subnets [14:14] i have 4 subnets. all on different vlans [14:14] on the default interface of the maas server the commissioning works [14:14] on all others it doesnt [14:15] but the interfaces are untagged all [14:15] im getting dhcp packets [14:16] which is nice ... the vm boots up but then cloud-init fails with data source missing [14:17] but i cannot read the output from the console because its too fast. there is no rsyslog logging yet because its in a too early stage [14:29] does fabric-0 untagged need to speak with fabric-1 untagged for example during the installation? [16:56] Bug #1673525 opened: Boot from SSD in AHCI mode fails [16:59] roaksoax, can you update the mailing list in the topic please - https://lists.ubuntu.com/mailman/listinfo/maas-devel [17:02] so i installed a second rack controller added the subnet there now commissioning works [17:02] the question is do i have to install a rack controller for every subnet now === frankban is now known as frankban|afk [20:08] Bug #1673598 opened: [2.1+] "boot_disk" for machine over API is return "null" [21:57] Hey, anyone around? [22:12] >_< [22:16] I'm here, but not likely helpfully so. (I have a wierd problem where I can't establish SSL with cloud-images.ubuntu.com unless I deploy it twice [22:17] I'm kinda starting to think SuperMicro just has lousy network cards [22:20] Oh weird.. [22:20] Yeah.. Something I did at some point seems to have changed to prevent me from releasing/erasing a node that was doing the lifecycle fine before.. [22:21] Haven't noticed anything like that with our SuperMicro machines, but we're using a big mix of different cards so I'm not even sure which ones are the onboards at the moment. [22:35] Yeah, nothing related to this seems to be working... Can't properly release, commission, or deploy.. [22:35] no useful error messages, other than maybe: provisioningserver.utils.services: [critical] mDNS observation process failed [22:35] But I can't find anything about it [comes up after the node has PXE booted from MAAS, and the node's console it filled with cloud-init messages about handlers.py[WARNING] failed posting event [22:38] derekcat: check that /etc/maas/rackd.conf is notpointint to localhost, but rather, an IP address the machine you deploying can reach to [22:39] roaksoax: yep, the maas_url is pointed at my VIP [23:14] I have a simple setup where my region controller is also my rack controller. It's currently telling me that it is "Missing connections to 1 region controller(s)." even though its the same machine [23:14] Is that likely related to what's going on with https://bugs.launchpad.net/maas/+bug/1660182 ? [23:15] Does it never attempt to reconnect? (Logs have to go back more than a day where there seems to be an issue.) [23:18] andrew-ii: i believe that's fixed in 2.2 [23:20] I'm on 2.1.3 - if I'm mostly just experimenting, would it be a good idea to switch to next? Or should I just ignore that error? [23:20] andrew-ii: restart maas-rackd and/or maas-regiond too if needed [23:21] andrew-ii: after the fix in 2.2 we have not seen the error so, we just need to make sure it is the case before we backport [23:21] roaksoax: Oh of course, one moment... [23:25] roaksoax: I should have realized that would fix it. `systemctl restart maas-rackd` did it. [23:35] cool