pmatulis | how do i deploy a node using the cli? | 02:31 |
---|---|---|
pmatulis | (v.1.8) | 02:31 |
mup | Bug #1518226 opened: Option to add ipmipower args via webUI <canonical-bootstack> <MAAS:New> <https://launchpad.net/bugs/1518226> | 09:46 |
caribou | Hello, is anyone familiar with how the booloading (BIOS/UEFI) is handled in Maas ? | 09:52 |
nagyz | hi guys | 10:32 |
nagyz | I'm trying to figure out what happens when I have a server on two MAAS-managed networks (one provisioning, one "public") | 10:32 |
nagyz | it seems to be that the default gateway will be set based on which interface is the first one in the alphabet :-) | 10:33 |
nagyz | same for DNS | 10:33 |
matt_dupre | Hi: I've just upgraded from 1.5 to 1.8.3, but I'm now seeing 503 errors trying to connect to http://<server>/MAAS/ | 12:17 |
matt_dupre | The apache error log says "Connection refused [...] attempt to connect to 127.0.0.1:5240 (localhost) failed". Nothing's listening on port 5240. Does anyone have any ideas what this could be? | 12:18 |
Haris | hello all | 13:45 |
Haris | what's maas for ubuntu ? | 13:45 |
Haris | is it something like .. if I needed running 2 or more VMs on the same machine, and I didn't have vmware, I'd use maas to do it ? | 13:45 |
Haris | os is it like a PXE ? | 13:47 |
roaksoax | Haris: MAAS is a deployment tool, that not only deploys Ubuntu but other operating systems | 13:54 |
Haris | so, its a PXE ? | 13:54 |
Haris | replacement ? | 13:54 |
roaksoax | Haris: it does PXE, UEFI, Grub net booting | 13:55 |
roaksoax | Haris: but it is not a "PXE replacement" | 13:55 |
roaksoax | Haris: PXE is an underlying technology | 13:55 |
Haris | I see | 13:56 |
Haris | is maas a web tool ? | 13:56 |
Haris | or just the front end | 13:56 |
roaksoax | Haris: maas provides a WebUI and a CLI, so you can do the same thing via those two interfaces | 13:59 |
roaksoax | Haris: so it is a software tool that incorporates other technologies (PXE, DHCP, DNS< Proxy), and provides users with a WebUI and a CLI Interface to control it | 14:00 |
Haris | ah ok | 14:00 |
Haris | cool | 14:00 |
mup | Bug #1518226 changed: Option to add ipmipower args via webUI <canonical-bootstack> <MAAS:Opinion> <https://launchpad.net/bugs/1518226> | 14:14 |
mup | Bug #1518226 opened: Option to add ipmipower args via webUI <canonical-bootstack> <MAAS:Opinion> <https://launchpad.net/bugs/1518226> | 14:20 |
Haris | thank you all | 14:31 |
mup | Bug #1518226 changed: Option to add ipmipower args via webUI <canonical-bootstack> <MAAS:Opinion> <https://launchpad.net/bugs/1518226> | 14:32 |
tinks | quit | 14:37 |
matt_dupre | So I figured by problem above out: /var/log/maas/maas.log was owned by syslog, but the user responsible for logging changed during the upgrade to maas, so maas-regiond was unable to write to the log | 15:08 |
matt_dupre | s/by/my/ | 15:09 |
los | Has anyone ever noticed that when you do dpkg-reconfigure on a cluster controller (without a regional) to point to a regional somewhere else....it might "ignore" your URL to the API and try the lo and another address 10.0.2.15? | 15:41 |
los | I can lynx the API and get a login from this VM, no problem, but the logs don't mention failing to connect to my URL I put into the reconfigure. | 15:42 |
los | ...and dpkg-reconfigure maas-cluster-controller takes a really long time to initially start and get into the CUI, if that's a clue | 15:43 |
los | The 10.0.2.15 is one of the local interfaces | 15:47 |
=== jfarschman is now known as MilesDenver | ||
roaksoax | los: do you have an examepl? | 15:55 |
roaksoax | los: or a pb ? | 15:55 |
los | roaksoax:pb? | 15:56 |
los | roaksoax: 2015-11-20 09:55:56-0600 [Uninitialized] Event-loop maasr6:pid=1101 (127.0.0.1:5251): Connection was refused by other side: 111: Connection refused. 2015-11-20 09:55:56-0600 [Uninitialized] Event-loop maasr6:pid=1101 (10.0.2.15:5251): Connection was refused by other side: 111: Connection refused. 2015-11-20 09:56:26-0600 [-] Event-loop maasr6:pid=1101 (192.168.66.1:5251): User timeout caused connection failure. | 15:57 |
los | Even though I've set the MAAS API URL to something totally different, its not in those clusterd.log 's | 15:57 |
los | roaksoax: Even though I can Lynx to that URL no problem, login, see stuff etc. from that same host | 15:58 |
los | roaksoax: the cluster-only host has that 10.0.2.15 address | 15:58 |
los | roaksoax: I read the logs as "I'm gonna connect to localhost on a few ports to see a regional, and I'll try my own IP" | 15:59 |
los | roaksoax: But I can't find that IP in any grep-able file | 15:59 |
roaksoax | los: /etc/maas/clusterd.conf is where the cluster knows where to contact the region controller | 16:00 |
roaksoax | .win 22 | 16:00 |
los | roaksoax: looking | 16:01 |
los | roaksoax: It has the right URL. I will bounce server. | 16:02 |
los | roaksoax: .win 22 = Old Yeller Headache? | 16:02 |
los | roaksoax: as one pal of my says, "Kevork it" | 16:02 |
roaksoax | los: so it has the right URL where to connect to MAAS Region ? | 16:06 |
los | roaksoax: yep | 16:06 |
roaksoax | los: did you add a shared secret ? | 16:06 |
los | roaksoax: yep, even the right one :) | 16:06 |
los | (this time) | 16:06 |
roaksoax | los: that's strange there, shows that it failed due to a timeout | 16:07 |
los | roaksoax: I thought the key part that was interesting is that the 10.13.x.x host I have the regional+cluster ("R+C1") is not mentioned. This is the 2nd attempted cluster controller ("C2") | 16:11 |
los | roaksoax: I'm assuming "if you can lynx it and get a login, you're good" | 16:11 |
los | roaksoax: Ohhhhhhh I bet the API doesn't work with the port specifier? | 16:13 |
los | roaksoax: (If that's the case the logging is bugged) | 16:13 |
roaksoax | los: what does your clusterd.conf show ? | 16:21 |
los | roaksoax: cluster_uuid: eea5c60a-4fcc-41a4-b141-30ca1b77c177 maas_url: http://10.13.0.6:8946/MAAS | 16:22 |
los | (two lines) | 16:22 |
roaksoax | http://10.13.0.6:8946/MAAS -> http://10.13.0.6:5240/MAAS | 16:23 |
los | roaksoax: clusterd.conf owned root, group maas, 640 | 16:23 |
roaksoax | los: or: http://10.13.0.6/MAAS | 16:23 |
los | roaksoax: 8946->80 | 16:26 |
los | roaksoax: Its in a VM under Vbox, I know, not supported, but its what I got :) | 16:26 |
roaksoax | los: right so the region binds to 5240, buyt apache is there to proxy from 80->5140 | 16:27 |
roaksoax | los: right so the region binds to 5240, buyt apache is there to proxy from 80->5240 | 16:27 |
los | roaksoax: ok, cool, so I'm forwarding in 8946 on the vmhost to the vm's 80...should I forward more? Is there some reverse DNS problem? Why is the clusterd not logging a connection attempt? (no sign it "sees" the MAAS API Url as set by dpkg-reconfigure) | 16:28 |
los | Is it a problem that the clusterd is otherwise not configured....aka, the 2nd interface on the C2 VM hasn't been setup yet | 16:30 |
roaksoax | los: my guess is that the forwarding is interfering here | 16:35 |
los | roaksoax: Can you give me an idea of the forwarding requirements? 5240->5240? | 16:36 |
roaksoax | los: yeah I'd do that | 16:37 |
los | roaksoax: using iptraf: can see local socket connection attempts to local ip 10.0.2.15 on ports 5250, 5253, now 5252 | 16:42 |
los | roaksoax: I have those forwarded on the R+C1 vm, but, that's the wrong IP. Its "ignoring my MAAS API" and trying local | 16:43 |
los | roaksoax: On the R+C1 vm host, I have 5250-3 forwarded into the R+C1 vm but no traffic on that | 16:43 |
mup | Bug #1518067 changed: multi-network node could get wrong default gateway that prevents deploy <MAAS:Invalid> <https://launchpad.net/bugs/1518067> | 16:44 |
los | roaksoax: This is interesting: | 16:46 |
los | │┌10.0.2.15:39872 = 5 300 S--- eth0 │ │└192.168.66.1:5251 = 0 0 ---- eth0 │ | 16:46 |
los | │└192.168.66.1:5251 = 0 0 ---- eth0 │ | 16:46 |
los | iptraf shows it making a connection from the NAT side network, to an address that I've used for things before but don't have in /etc/network/interfaces at all | 16:47 |
los | roaksoax: I have eth0, enp0s8, and lo....s8 is totally unconfigured. | 16:48 |
los | roaksoax: Ok, I configured the enp0s8 network (That will be the private MaaS network) now I'm getting the connection refused msgs 1/sec, but just the two last ones, the 'refused by other side' | 16:51 |
los | roaksoax: The third (first showing) message with "Timeout" is gone | 16:51 |
los | roaksoax: And...now the dpkg-reconfigure maas-cluster-configure is not slow to start! Aha! | 16:53 |
los | roaksoax: Changed the URL to 10.13.0.6/MAAS and now I get a fast fail from twisted.... 404 | 16:53 |
los | roaksoax: Aha, totally different clusterd.log, after changing the URL to 10.13.0.6:5240/MAAS and having the 5240->5240 mapping | 16:55 |
los | 2015-11-19 13:41:08-0600 [-] Site starting on 5248 2015-11-19 13:41:08-0600 [-] TFTP starting on 69 2015-11-19 13:41:08-0600 [-] Starting protocol <tftp.protocol.TFTP instance at 0x7f06b03b8998> 2015-11-19 13:41:08-0600 [-] TFTP Listener started at 127.0.0.1:69 2015-11-19 13:41:08-0600 [-] TFTP starting on 69 2015-11-19 13:41:08-0600 [-] Starting protocol <tftp.protocol.TFTP instance at 0x7f06b03b8cf8> 2015-11-19 13:41:08-0600 [-] TFTP Lis | 16:55 |
los | 2015-11-19 13:41:08-0600 [Uninitialized] Region not available: Connection was refused by other side: 111: Connection refused. (While requesting RPC info at http://localhost:5240/MAAS/rpc/) | 16:56 |
los | roaksoax: iptraf now shows its reaching out to the right 10.13.0.6! Closer! | 16:56 |
los | roaksoax: On the R+C1 machine, Apache says: Nov 20 11:03:39 maasr6 sh[1091]: 2015-11-20 11:03:39 [-] 127.0.0.1 - - [20/Nov/2015:17:03:38 +0000] "GET /MAAS/rpc/ HTTP/1.0" 2 00 364 "-" "provisioningserver.rpc.clusterservice.ClusterClientService" | 17:04 |
los | roaksoax: 2015-11-20 11:06:29-0600 [Uninitialized] Region not available: Connection was refused by other side: 111: Connection refused. (While requesting RPC info at http://10.13.0.6:5240/MAAS/rpc/). | 17:06 |
los | Better...but not good | 17:06 |
roaksoax | los: right, so that means the cluster, for whatever reason, can't reach to the region | 17:08 |
los | roaksoax: On the R+C1 machine, when I curl localhost:5240 I get <html> <head> <meta http-equiv="refresh" content="0;URL=MAAS/"> </head> <body bgcolor="#FFFFFF" text="#000000"> <a href="MAAS/">click here</a> </body> </html> | 17:08 |
roaksoax | los: might be the forwarding | 17:08 |
los | roaksoax: On C2 machine I get ConnRefused so I'm baffled but on it | 17:08 |
los | For the good of all MaaSKind, lolz | 17:08 |
roaksoax | los: this is what happens, the cluster connects to region on that IP/port, and the Region tells the cluster "this are my 4 endpoints, on different ports, because I have multiple daemons running" | 17:09 |
roaksoax | los: so that's the problem I'd think | 17:09 |
roaksoax | los: the cluster cannot connect to the endpoints | 17:09 |
los | roaksoax: Do I need to forward a range of ports, 5240-X -> 5240-X? | 17:09 |
roaksoax | los: "Firewall ports for Region and Cluster controller communication | 17:13 |
roaksoax | The communication between Region and Cluster controller is now limited to use the ports between 5250 and 5259. For all of those users who are using a remote cluster (not running on the same machine as the MAAS Region Controller), need to ensure that these ports are open in the firewall." | 17:13 |
los | roaksoax: Aha, forsooth! | 17:13 |
los | roaksoax: Ok, curl to the 5240 is getting me the "click here" html" | 17:16 |
los | roaksoax: Ok, I get this msg 3x: 2015-11-20 11:26:04-0600 [Uninitialized] Event-loop maasr6:pid=1096 (127.0.0.1:5252): Connection was refused | 17:26 |
roaksoax | uhmm | 17:27 |
los | roaksoax: ...but the IP changes: I get back a 10.0.2.15, which is (I'm thinking) dispensed by the regional, which is an unroutable | 17:27 |
los | roaksoax: So, "Oh, I contact the regional, and it gave me an address and I'm gonna connect and its no good, try these other interfaces address he gave me and loopback too" | 17:27 |
roaksoax | bladernr_: ^^ | 17:27 |
roaksoax | err | 17:27 |
roaksoax | blake_r: ^^ | 17:27 |
los | roaksoax: I appreciate the help: I recognize this is all virtualbox hooey | 17:28 |
roaksoax | blake_r: any ideas? seems that region might be sending back loopback to cluster ? | 17:28 |
los | roaksoak: I changed the local private maas nets on both the R+C1 machine and C2 machines to non-192.168.66.x addresses (.44 and .88) and then it went bonkers. Log ceased...a couple timeouts...long pauses...so I'm bouncing both hosts | 17:32 |
los | roaksoax: Ok, this is cool: the .88 address is showing up in the C2 machine's logs, so R+C1 machine is dispensing it back to the C2 machine, ergo, this is all a consequence of the virtual box and unroutables | 17:35 |
los | roaksoax: Clear as a day, the Regional is passing back ALL the interfaces addresses that are non-loopback, and then the Cluster controller is trying them all....I see iptraf on the C machine attempting to connect out to 192.168.88.1, which is the R+C1 machines private MaaS network for the "dummies" to boot off. | 17:36 |
los | of | 17:36 |
los | Roaksoax: taking lunch. Will work this more when I get back. Will be looking at netstat to see what's listening on the R+C1 box, etc. | 17:43 |
mup | Bug #1518414 opened: NVMe model info/serial number is not seen <MAAS:New> <https://launchpad.net/bugs/1518414> | 18:12 |
mup | Bug #1518414 changed: NVMe model info/serial number is not seen <MAAS:New> <https://launchpad.net/bugs/1518414> | 18:15 |
mup | Bug #1518414 opened: NVMe model info/serial number is not seen <MAAS:New> <https://launchpad.net/bugs/1518414> | 18:18 |
los | roaksoax: So my region controller is just happy as can be: 2015-11-20 13:35:03 [-] 127.0.0.1 - - [20/Nov/2015:19:35:02 +0000] "GET /MAAS/rpc/ HTTP/1.0" 200 364 "-" "provisioningserver.rpc.clusterservice.ClusterClientService" | 19:35 |
los | roaksoax: should that be on the loopback? | 19:36 |
los | roaksoax: hahahaha the R+C1 machine has localhost:5240 in its cluster controller config! | 19:40 |
los | roaksoax: Wow: I provisioned the R+C1 machine with two interfaces, and I config'd the maas-region-controller with that (NAT outbound) interface up and the other one (MaaS private) unconfigured and its my bad...MaaS picked the only up network as the private one | 19:44 |
los | roaksoax: I get one of these when I trigger a restart through dpkg: 2015-11-20 13:53:20-0600 [HTTPPageGetter,client] Region is not advertising RPC endpoints. | 19:54 |
mup | Bug #1518440 opened: tgt fails to install in LXD <MAAS:Triaged> <tgt:New> <tgt (Ubuntu):New> <https://launchpad.net/bugs/1518440> | 20:06 |
mup | Bug #1518440 changed: tgt fails to install in LXD <MAAS:Triaged> <tgt:New> <tgt (Ubuntu):New> <https://launchpad.net/bugs/1518440> | 20:09 |
los | roaksoax: Doc bug: right above this link, the sentence just.... https://maas.ubuntu.com/docs/install.html#disc-install | 20:10 |
mup | Bug #1518440 opened: tgt fails to install in LXD <MAAS:Triaged> <tgt:New> <tgt (Ubuntu):New> <https://launchpad.net/bugs/1518440> | 20:12 |
mup | Bug #1518440 changed: tgt fails to install in LXD <MAAS:Triaged> <tgt:New> <tgt (Ubuntu):New> <https://launchpad.net/bugs/1518440> | 20:15 |
mup | Bug #1518440 opened: tgt fails to install in LXD <MAAS:Triaged> <tgt:New> <tgt (Ubuntu):New> <https://launchpad.net/bugs/1518440> | 20:18 |
nagyz | can someone enlighten me how would you manage DNS if you have two different subnets, like an internal and an external? | 22:33 |
nagyz | if a node has IPs on both... currently MAAS only registers the A record for the alphabetically first interface | 22:34 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!