| *** sonali5 has quit IRC | 00:04 | |
| *** terriyu has quit IRC | 01:22 | |
| *** sonali5 has joined #openstack-opw | 04:01 | |
| *** rossella_s has quit IRC | 04:30 | |
| *** rossella_s has joined #openstack-opw | 04:31 | |
| *** GB21 has joined #openstack-opw | 04:46 | |
| *** ig0r__ has joined #openstack-opw | 05:15 | |
| *** ig0r_ has quit IRC | 05:19 | |
| *** ig0r__ has quit IRC | 05:21 | |
| *** enthurohini has joined #openstack-opw | 07:17 | |
| *** enthurohini has quit IRC | 07:19 | |
| *** GB21 has quit IRC | 07:27 | |
| *** ig0r_ has joined #openstack-opw | 07:32 | |
| *** GB21 has joined #openstack-opw | 07:43 | |
| *** ig0r_ has quit IRC | 08:11 | |
| *** ig0r_ has joined #openstack-opw | 08:19 | |
| *** rossella_s has quit IRC | 08:30 | |
| *** rossella_s has joined #openstack-opw | 08:30 | |
| *** ig0r_ has quit IRC | 08:53 | |
| *** ig0r_ has joined #openstack-opw | 09:19 | |
| *** Zyric1 has quit IRC | 09:24 | |
| *** ig0r_ has quit IRC | 09:47 | |
| *** neiljerram has joined #openstack-opw | 09:48 | |
| *** ig0r_ has joined #openstack-opw | 11:08 | |
| *** aysyd has joined #openstack-opw | 11:16 | |
| *** aysyd has quit IRC | 11:20 | |
| *** aysyd has joined #openstack-opw | 11:20 | |
| *** ig0r_ has quit IRC | 11:46 | |
| *** GB21 has quit IRC | 11:51 | |
| *** julim has joined #openstack-opw | 12:06 | |
| *** psargent has quit IRC | 12:06 | |
| *** psargent has joined #openstack-opw | 12:07 | |
| *** ig0r_ has joined #openstack-opw | 12:15 | |
| *** ig0r_ has quit IRC | 12:24 | |
| *** _amrith_ is now known as amrith | 12:31 | |
| *** njirap has joined #openstack-opw | 12:58 | |
| *** ig0r_ has joined #openstack-opw | 13:16 | |
| *** ig0r_ has quit IRC | 13:25 | |
| *** lblanchard has joined #openstack-opw | 13:30 | |
| *** amrith is now known as _amrith_ | 13:52 | |
| *** GB21 has joined #openstack-opw | 14:06 | |
| *** ig0r_ has joined #openstack-opw | 14:41 | |
| *** ig0r_ has quit IRC | 14:43 | |
| *** _amrith_ is now known as amrith | 14:49 | |
| *** enthurohini has joined #openstack-opw | 15:01 | |
| psargent | neiljerram: are you there? | 15:01 |
|---|---|---|
| *** ig0r_ has joined #openstack-opw | 15:01 | |
| neiljerram | psargent, hi, yes! | 15:01 |
| psargent | good afternoon! | 15:02 |
| neiljerram | Good afternoon, and thanks for your progress report. | 15:02 |
| psargent | Did you see my e-mail? | 15:02 |
| psargent | ok..you got it | 15:02 |
| neiljerram | Yes, and I have a few comments... | 15:02 |
| psargent | ok..great! | 15:02 |
| neiljerram | Firstly, it's quite normal to have to unstack and rerun DevStack again, after making some kind of config change. | 15:03 |
| neiljerram | (Even though it takes time, which is annoying...) | 15:03 |
| psargent | ok!! I had to unstack it several times! | 15:04 |
| neiljerram | Sometimes it can be even worse, in the sense that a previous set of config leaves something behind that still causes a theoretically correct new config to fail. | 15:04 |
| vkmc | psargent, hi there! suggestion about how to use IRC... if you see someone online and you want to tell this person something, mention them and leave the message, so they can get back to you when they can | 15:04 |
| psargent | vkmc: ok thanks for the suggestion! | 15:05 |
| neiljerram | So far that reason people open do DevStack on a VM where they've taken a snapshot just after installing the OS; so that they can easily revert to that snapshot and then have confidence that DevStack will not depend on prior history. | 15:06 |
| *** ig0r_ has quit IRC | 15:06 | |
| neiljerram | But I don't know if you have access to VMs... | 15:07 |
| neiljerram | You also asked: What code is running when I created the network and created the VMs? Is there a way I can trace the code ? | 15:07 |
| psargent | I haven't been taking a snapshot but I can see how that is very helpful!!! | 15:08 |
| neiljerram | When you create the network, code runs in the Neutron server (q-svc.log) and creates objects in the Neutron DB that describe that network. | 15:09 |
| psargent | I'm not sure how to take a snapshot of the VM. | 15:09 |
| psargent | ok...so I'll take a look at q-svc.log... | 15:10 |
| *** sonali5 has quit IRC | 15:10 | |
| neiljerram | It's possible that messages are also sent to the Neutron agent code at that time, but I'm not sure. | 15:11 |
| neiljerram | (If you have separate controller and compute nodes, the server is what runs on the controller, and the agent is what runs on the compute nodes.) | 15:12 |
| neiljerram | When you create a VM, a lot happens, mostly involving Nova and Neutron. | 15:12 |
| psargent | ok...I understand | 15:13 |
| neiljerram | The 'nova boot' command, or Horizon UI, sends a request to Nova to boot a VM, including the ID of the network that the VM should be attached to. | 15:13 |
| neiljerram | Nova then sends a request to Neutron, to ask Neutron to create a 'port' for that VM on the specified network. | 15:14 |
| psargent | ok...I am following | 15:14 |
| psargent | is the Horizon UI the same thing as the dashboard? | 15:16 |
| neiljerram | When Neutron gets that request, lots of things happen, including the allocation of an IP address - from the subnets that are associated with the network - and calls into the Neutron 'plugin' so that it can do any setup that is needed for that port. | 15:16 |
| neiljerram | Yes, Horizon UI = dashboard | 15:16 |
| psargent | ok | 15:16 |
| neiljerram | Usually the 'plugin' is the 'ML2' plugin, as this is now the reference plugin that core Neutron provides. But in general it can be something deployment-specific. | 15:17 |
| psargent | is the ML2 plugin something that is setup in the local.conf file? | 15:19 |
| neiljerram | Assuming all that port setup succeeds, Neutron sends a response back to Nova with the details of the port that Nova needs for its VM-launching code - for example a corresponding Linux interface name. | 15:19 |
| neiljerram | Then Nova can actually boot the VM - which it usually does using a (non-OpenStack) library called libvirt. | 15:20 |
| *** GB21 has quit IRC | 15:20 | |
| neiljerram | Nova creates an XML config that describes the VM properties, including some details of the port into it, and then tells libvirt to 'make it so' with that XML config | 15:21 |
| neiljerram | Now the ML2 plugin also has things called 'type' and 'mechanism' drivers, which handle the detail of how data sent by a VM (such as a ping) gets to any other VM. | 15:23 |
| neiljerram | In your DevStack setup, I guess you are using the openvswitch mechanism driver, as that is the default. | 15:24 |
| psargent | yes...I am using the default | 15:24 |
| neiljerram | (You could check this in a file named ml2_conf.ini, which lives somewhere under /etc/neutron/.) | 15:26 |
| psargent | ok...i'll check | 15:26 |
| neiljerram | When one of your VMs pings the other one, it sends the ping request out of its eth0 interface. | 15:27 |
| *** njirap has quit IRC | 15:28 | |
| neiljerram | The same packet then appears on the VM's host on an interface named something like tap12345-abcd. | 15:28 |
| psargent | ok....I issued a command called ifconfig...and think I saw something about eth0 interface | 15:29 |
| neiljerram | The tap12345-abcd interface is plugged into an openvswitch bridge, as also is the similar tap... interface that connects to your other VM. | 15:29 |
| neiljerram | So basically the ping packet comes down tap... interface, through the bridge, and out the other tap... interface. And then the other VM receives the request. | 15:30 |
| neiljerram | Sorry, probably a bit more answer than you were asking for, there.... | 15:31 |
| psargent | this is great information for me ! | 15:32 |
| neiljerram | You also asked if you can trace the code. | 15:32 |
| enthurohini | neiljerram, for other interns also. thanks | 15:33 |
| neiljerram | Depends what you mean exactly by tracing. | 15:33 |
| neiljerram | All of the components of OpenStack generate log files, and write various more-or-less useful traces into those. | 15:34 |
| neiljerram | In a DevStack setup, you can find these at /opt/stack/logs. In other setups they are usually at /var/log/nova, /var/log/neutron, etc. | 15:34 |
| neiljerram | But you might also breaking into the Python code, like with a debugger, and stepping through the code that is being executed. | 15:35 |
| psargent | I was looking for a trace so I can see the flow of the procedure calls | 15:35 |
| neiljerram | I think that can probably be done too, using the python debugger (pdb), but I'm afraid I don't know how. | 15:35 |
| neiljerram | OK, for that, I think the first place is to look at the log files. | 15:36 |
| neiljerram | It can be tricky, though, because there is no overall view that shows you all the components at once. | 15:36 |
| neiljerram | (as far as I know, anyway) | 15:37 |
| neiljerram | You have to look at the Neutron server log (q-svc), to see what it thinks, and then at the DHCP agent log (q-agt) to see what it did, and then at Nova (n-...) to see its point of view, and so on. | 15:38 |
| neiljerram | (I recently tried to start working on a tool that would give a better overall picture, but I haven't got very far with that yet - see http://lists.openstack.org/pipermail/openstack-dev/2015-November/079571.html) | 15:40 |
| psargent | ok....I see...I wanted to get an idea of when I use the dashboard to created a VM what cod gets executed | 15:40 |
| psargent | oh ok ..cool! | 15:40 |
| neiljerram | But anyway, great that you have a working DevStack system now! | 15:41 |
| *** njirap has joined #openstack-opw | 15:41 | |
| psargent | I wanted to get an idea of when I use the dashboard (or CLI) to create a VM what code gets executed... | 15:42 |
| neiljerram | Yes, indeed, that's important. | 15:42 |
| neiljerram | Also good to know is that you can make more information go into the logs, by changing the log level, e.g. to DEBUG. | 15:43 |
| *** Akanksha08 has joined #openstack-opw | 15:43 | |
| psargent | ok... | 15:43 |
| neiljerram | There are two settings in /etc/neutron/neutron.conf: debug and verbose. | 15:43 |
| neiljerram | If you set both of those to True, you'll get a lot more information in the Neutron log files. | 15:44 |
| psargent | ok...I will set both to True and check out the logs | 15:44 |
| neiljerram | That will include the file names and line numbers of the code that is being executed, so then you can go and match the log files to the relevant code. | 15:45 |
| psargent | Great! That is what I am looking for! | 15:45 |
| neiljerram | After changing neutron.conf, you'll need to restart the relevant component. Normally that's done by, e.g., 'service neutron-server restart' (for the Neutron server), but in DevStack it's different... | 15:46 |
| neiljerram | Have you seen ./rejoin-stack.sh yet? | 15:46 |
| psargent | no...I haven't seen ./rejoin-stack.sh | 15:47 |
| neiljerram | OK, so after you do ./stack.sh, you can also do ./rejoin-stack.sh. Actually this is another way of seeing all the logs. | 15:49 |
| neiljerram | It's a bit tricky to describe... | 15:49 |
| psargent | ok...i'll check it out | 15:49 |
| psargent | my next steps are to set up the logs (debug and verbose) and then create a VM and see what code gets executed. Does that sound ok or should I focus on something else? | 15:51 |
| neiljerram | It gives you a GNU Screen session with a logical terminal for every OpenStack component that is running. | 15:51 |
| psargent | ok..nice and then I can see all the logs too | 15:52 |
| neiljerram | So there will be a terminal for the Neutron server, and the way to restart it would be: switch to that terminal, Ctrl-C to kill the server, Up arrow to repeat the previous command (which starts the server running again), and Enter. | 15:52 |
| neiljerram | Yes, that sounds good. And as part of that, try to become familiar with rejoin-stack.sh too. | 15:53 |
| neiljerram | (As what I've said may not have made much sense.) | 15:53 |
| neiljerram | And obviously, a particular thing that you may want to keep in mind is: where and how does the IP address get allocated? | 15:54 |
| psargent | ok...I'll definitely check out rejoin-stack | 15:54 |
| psargent | ohh..yes! | 15:54 |
| psargent | I'll keep that in mind too | 15:56 |
| neiljerram | That's possibly enough for today... shall we plan to speak again on Monday? | 15:56 |
| psargent | Yes, let's plan for Monday....same time...same channel! | 15:57 |
| neiljerram | OK, cool - see you then! | 15:57 |
| psargent | ok..have a great weekend! | 15:57 |
| *** ig0r_ has joined #openstack-opw | 16:21 | |
| *** rossella_s has quit IRC | 16:30 | |
| *** GB21 has joined #openstack-opw | 16:30 | |
| *** rossella_s has joined #openstack-opw | 16:30 | |
| *** GB21 has quit IRC | 16:34 | |
| *** terriyu has joined #openstack-opw | 16:42 | |
| *** terriyu has quit IRC | 16:43 | |
| *** terriyu has joined #openstack-opw | 16:43 | |
| *** terriyu has quit IRC | 16:45 | |
| *** terriyu has joined #openstack-opw | 16:46 | |
| *** ig0r_ has quit IRC | 16:47 | |
| *** GB21 has joined #openstack-opw | 16:51 | |
| *** sonali5 has joined #openstack-opw | 17:17 | |
| *** terriyu has quit IRC | 17:44 | |
| *** njirap has quit IRC | 18:04 | |
| *** sonali5 has quit IRC | 18:09 | |
| *** njirap has joined #openstack-opw | 19:00 | |
| *** neiljerram has quit IRC | 19:00 | |
| *** aysyd has quit IRC | 19:12 | |
| *** aysyd has joined #openstack-opw | 19:13 | |
| *** Akanksha08 has quit IRC | 19:19 | |
| *** amrith is now known as _amrith_ | 20:08 | |
| *** Zyric has joined #openstack-opw | 20:10 | |
| *** njirap has quit IRC | 20:37 | |
| *** ig0r_ has joined #openstack-opw | 20:37 | |
| *** njirap_ has joined #openstack-opw | 20:37 | |
| *** GB21 has quit IRC | 20:54 | |
| *** aysyd has quit IRC | 21:09 | |
| *** aysyd has joined #openstack-opw | 21:12 | |
| *** _amrith_ is now known as amrith | 21:24 | |
| *** psargent has quit IRC | 21:26 | |
| *** lblanchard has quit IRC | 22:04 | |
| *** aysyd has quit IRC | 22:08 | |
| *** enthurohini has quit IRC | 23:14 | |
| *** njirap_ has quit IRC | 23:39 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!