(05:02:20 PM) jmelis: Hello everyone we're going to start answering the questions
(05:03:04 PM) jmelis: ask anything related with this meeting in channel #opennebula-questions
(05:03:09 PM) jmelis: the first question is:
(05:03:31 PM) jmelis: vbernetr: Are there any plans to support hotplug on a VM, be it attaching a block device, a vif, more CPUs, or extra RAM ? How hard would it be to integrate this with the existing core ?
(05:04:45 PM) jfontan: There are already interest in at least attach block devices in running VM's
(05:04:51 PM) jfontan: for example this ticket http://dev.opennebula.org/issues/304
(05:05:05 PM) rsmontero [~ruben@pc-ruben.dacya.ucm.es] entered the room.
(05:05:23 PM) jfontan: and this other one to support EBS like block devices http://dev.opennebula.org/issues/378
(05:05:39 PM) mode (+o rsmontero) by jmelis
(05:05:48 PM) jfontan: We cannot give you a timeframe on when this will be supported
(05:06:11 PM) jfontan: or if this will come with RAM, network or cpu hot modification
(05:06:14 PM) koushik [~koushik@tko-dhcp-75.cs.hut.fi] entered the room.
(05:06:25 PM) jfontan: but it certainly wont be for the next version
(05:07:15 PM) jmelis: does anyone have any doubts with this question?
(05:07:33 PM) vbernetr: thanks
(05:07:57 PM) jmelis: ok, let's go on to the next question
(05:08:13 PM) jmelis: vbernetr: the lifecycle system seems pretty tight and is basically a static system. Are there any plans for extending it, to make it hackable (for instance allowing power users to add other states/transitions ?
(05:08:15 PM) koushik is now known as uss
(05:08:39 PM) vvuksan [~vuksan@157.130.20.218] entered the room.
(05:09:29 PM) cmartin: well, the LCM is a core component. As you pointed out, it is a static system, and can't be modified with plug-ins or hooks
(05:10:09 PM) cmartin: but we don't have any plans to offer any functionality that would allow you to modify it dinamically
(05:10:38 PM) cmartin: on the other hand, the most difficult part is understanding the life-cycle itself
(05:11:26 PM) cmartin: once you got it, the C++ code can be modified "easily" for your needs
(05:11:33 PM) vvuksan: how would I go about changing the deployment.0 template ?
(05:11:43 PM) vvuksan: ie. I would like to add console.log
(05:11:48 PM) jmelis: vvuksan: please ask your questions in #opennebula-questions
(05:11:54 PM) vvuksan: k
(05:12:27 PM) cmartin: vbernetr, any comments?
(05:12:35 PM) vbernetr: cmartin: the thing is, maintening an fork of the core is harder than maintaining a plugin.
(05:13:09 PM) zenet [~zenet@246.39.103-84.rev.gaoland.net] entered the room.
(05:13:12 PM) vbernetr: Basically, I would like to be able to totally shut down a VM, and restart it later. For now, this means I have to keep some state outside of open nebula.
(05:13:24 PM) vbernetr: and create a new vm
(05:14:02 PM) cmartin: ok, those two operations seem reasonable
(05:14:03 PM) vbernetr: would it be possible to perhaps get a "reset" action, that would take a VM from any state to "pending", or something like that ?
(05:15:14 PM) cmartin: if you have any LCM feature request, we would be happy to open tickets in our dev portal
(05:15:30 PM) vbernetr: ok, I will do that :)
(05:15:45 PM) cmartin: the reset action you ask about is actually implemented, let me check the doc...
(05:16:08 PM) vbernetr: iirc, it doesn't work when a VM is "done"
(05:16:35 PM) cmartin: "onevm resubmit" is the command
(05:16:41 PM) cmartin: http://opennebula.org/_media/documentation:rel2.2:states-complete.png
(05:17:19 PM) cmartin: oh, yeah that's right. A VM in DONE state is considered as ended, you won't use it anymore
(05:18:03 PM) cmartin: the way to resubmit a DONE Vm would be to create another one from the same template, using the Images of the previous VM
(05:18:10 PM) cmartin: this will be easier in the next version
(05:18:16 PM) vbernetr: yeah, that's what I do for now
(05:18:18 PM) cmartin: as you will be able to register the templates
(05:18:27 PM) vbernetr: but as I said, it means keeping state outside
(05:18:30 PM) vbernetr: ah, interesting
(05:18:33 PM) cmartin: and just instantiate them as many times as you want
(05:18:43 PM) vbernetr: I like the idea :)
(05:19:07 PM) vbernetr: that would be for 2.4 ?
(05:19:14 PM) uss is now known as koushik
(05:19:34 PM) cmartin: anyway, we found the reboot ticket: http://dev.opennebula.org/issues/132
(05:20:28 PM) cmartin: you mean the template repository for 2.4?
(05:20:33 PM) vbernetr: yeah
(05:20:34 PM) cmartin: yes, it will be included
(05:20:37 PM) vbernetr: great :)
(05:21:00 PM) jmelis: ok, let's move on to the next question
(05:21:15 PM) jmelis: sneumann: dynamic re-scheduling of VMs. where running VMs are chosen (e.g. taken from the most busy node) and the (live-)migration target is determined by the scheduler. This would also allow to disable a node, and "empty" it, by migrating its VMs away to other nodes
(05:21:55 PM) jfontan: OpenNebula scheduler is a separate component
(05:22:07 PM) jfontan: and in reality more than one scheduler can coexist
(05:22:43 PM) jfontan: some time ago we've been playing with that idea and we had a rescheduler doing something similar
(05:23:37 PM) jfontan: the idea was to pack the VMs in the least of physical nodes and sending the shutdown command to the physical nodes that were not used
(05:23:48 PM) sneumann: exactly my intent
(05:24:20 PM) ugarit left the room (quit: Ping timeout: 252 seconds).
(05:24:22 PM) jfontan: the same way when VMs are in pending network wake up was issued to start new physical nodes and cope with the load
(05:24:50 PM) koushik left the room (quit: Remote host closed the connection).
(05:24:57 PM) jfontan: this is not a feature we are thinking on implementing in the OpenNebula distribution
(05:25:11 PM) jfontan: but it could be done as a ecosystem project
(05:25:24 PM) jfontan: as it is not really dependent on the core
(05:25:44 PM) koushik [~koushik@tko-dhcp-75.cs.hut.fi] entered the room.
(05:26:02 PM) jfontan: any comment?
(05:26:34 PM) sneumann: Is there an API desription for sched ?
(05:26:54 PM) sneumann: I'd need to ask it for the "best" target
(05:26:55 PM) jfontan: the API the shceduler uses is XML-RPC
(05:27:05 PM) jfontan: it basically does tha same as one* commands
(05:27:13 PM) sneumann: ack
(05:27:19 PM) jfontan: something like onevm list and get pending VM's
(05:27:28 PM) jfontan: and onehost list to get the resources
(05:27:37 PM) jfontan: then mix'n'match and issue onevm deploy
(05:28:14 PM) sneumann: more onevm live-migrate
(05:28:37 PM) jfontan: that will be the rescheduler
(05:28:38 PM) jfontan: yup
(05:29:12 PM) jmelis: sneumann: should we move on to the next question?
(05:29:21 PM) sneumann: yup
(05:29:23 PM) jmelis: ok
(05:29:28 PM) jmelis: next question is:
(05:29:31 PM) jmelis: sneumann: possibly a dynamic CPU determination, based on e.g. the 1,5 or 15mins load of a VM which is then considered by the scheduler. There might be more advanced ways to determine the actual/current/dynamic CPU value...
(05:31:04 PM) jfontan: at least for this or the next version it wont be possible
(05:31:25 PM) jfontan: to change that value, the CPU reserved for a VM
(05:31:32 PM) jfontan: there are two values that should be changed
(05:31:45 PM) jfontan: one is the VM template CPU value, that is the easy part
(05:32:00 PM) jfontan: and then the CPU available in the host the VM is running
(05:32:13 PM) jfontan: that is the "acpu" value you get with onehost list
(05:32:50 PM) jfontan: I think now it is a good time to add this ticket as we are changing the database for the next OpenNebula versions
(05:33:00 PM) jfontan: so it is easier to add some features
(05:33:16 PM) jfontan: and will be good to take those ideas into account
(05:33:42 PM) jfontan: any question?
(05:34:02 PM) sneumann: no
(05:34:24 PM) jfontan: it would be nice that you open that ticket
(05:34:53 PM) jmelis: ok, next question is:
(05:34:55 PM) jmelis: vbernetr: The polling of VMs with Xen 4.0.1 is broken due to xentop output format change. How can I help fix that in an adequate fashion, and get it commited upstream ?
(05:36:42 PM) jfontan: we are aware of your ticket http://dev.opennebula.org/issues/567
(05:37:04 PM) jfontan: and we are checking the best way to make it compatible with both versions
(05:37:46 PM) deshantm_laptop left the room (quit: Quit: ChatZilla 0.9.86.1 [Firefox 4.0.1/20110422203143]).
(05:37:48 PM) jfontan: if this won't be possible we will provide probes for both versions but most probably we will just check for the version and execute the correct codepath
(05:38:08 PM) jfontan: we expect to have that working for 2.4
(05:38:28 PM) jfontan: but most probably this patches will be also working for 2.2
(05:38:38 PM) vbernetr: ok. I have time for this, are you interrested in a patch with several code paths ?
(05:39:13 PM) jfontan: sure
(05:39:20 PM) jfontan: that would be great
(05:39:35 PM) jfontan: as we must be installing 4.0 here to do all those tests
(05:40:09 PM) vbernetr: Debian has 4.0.1 packages which is neat
(05:40:39 PM) vbernetr: I should have a patch ready within a few days.
(05:40:53 PM) sneumann left the room (quit: Ping timeout: 240 seconds).
(05:40:55 PM) jfontan: taking note and giving the info to our infrastructure manager
(05:41:06 PM) jfontan: thanks
(05:41:20 PM) jmelis: ok, let's move on to the next question
(05:41:27 PM) jmelis: ugarit: are there plans to virtualize GPUs?
(05:42:22 PM) jfontan: the idea of GPU virtualization is more a feature of hypervisor/OS configuration that OpenNebula itself
(05:42:33 PM) jfontan: we don't have experience with this
(05:43:07 PM) jfontan: but reading in the internet it seems that people chnage GPU(OpenGL/OpenCL) libraries with proxies
(05:43:36 PM) jfontan: that connect to a server in the physical nodes that does the real communication with the GPU
(05:44:07 PM) jfontan: if that is the case then OpenNebula should not be messing with that but the VMs should be prepared for that
(05:44:27 PM) jfontan: or some special configuration in the deployment files should be passed to the hypervisor
(05:44:39 PM) jfontan: maybe you have more information on this
(05:45:01 PM) jmelis: ugarit disconnected a while back
(05:45:05 PM) vbernetr: jfontan: another way would be PCI passthrough
(05:45:06 PM) jfontan: ops
(05:45:31 PM) jfontan: vbernetr: wont that make the device only available to a VM?
(05:45:53 PM) jfontan: as some kind of arbitration should be made
(05:46:38 PM) jmelis: ok, next question:
(05:46:51 PM) vbernetr: jfontan: indeed, which is good enough for GPU compute clusters.
(05:47:00 PM) jmelis: oops, sorry guys, go on
(05:47:19 PM) jfontan: I think it is a nice issue to add to the development pages and have a discussion on it
(05:47:37 PM) jfontan: as GPU computation is being used more and more these days
(05:48:31 PM) jmelis: ok, moving on:
(05:48:35 PM) jmelis: ugarit: are there plans to utilize openstack's object store capabilities?
(05:48:51 PM) ffeldhaus left the room (quit: Ping timeout: 252 seconds).
(05:49:11 PM) dmolina: We are developing a new tool for importing images from external storage providers
(05:49:24 PM) dmolina: Currently, we are finishing the S3 driver
(05:49:42 PM) dmolina: therefore you will be able to import your EC2 amis to OpenNebula
(05:50:06 PM) dmolina: other drivers can be implemented to support different providers
(05:51:19 PM) dmolina: ugarit, you can open open a ticket and we will consider to implement a new driver to support Swift
(05:52:07 PM) dmolina: furthermore in the next version the ImageRepository is implemented based on drivers
(05:52:24 PM) dmolina: so that you can implement a driver for it
(05:53:35 PM) dmolina: you can check the implementations that have been done in the following links:
(05:53:46 PM) dmolina: http://opennebula.org/software:ecosystem:moosefs
(05:54:38 PM) dmolina: http://mperedim.wordpress.com/2010/09/26/opennebula-zfs-and-xen-%E2%80%93-part-2-instant-cloning/
(05:55:08 PM) jmelis: ok, next question is:
(05:55:11 PM) jmelis: vvuksan: how would I go about changing the deployment.0 template ?
(05:55:33 PM) jmelis: the deployment.0 is generated by the core, so it's really not that easy to change it
(05:55:41 PM) vvuksan: :-(
(05:55:41 PM) jmelis: let me paste you a few files you might want to look at
(05:56:22 PM) jmelis: http://dev.opennebula.org/projects/opennebula/repository/revisions/master/entry/src/vmm/LibVirtDriverKVM.cc
(05:56:24 PM) jmelis: but
(05:56:49 PM) jmelis: there's an easier way. In oned.conf you can specify the deployment.0 to be an xml
(05:57:36 PM) jmelis: under the VM_MAD =
(05:57:37 PM) vvuksan: yikes
(05:58:02 PM) jmelis: simply change type="xen" or type = "kvm" to type ="xml"
(05:58:11 PM) vvuksan: have you guys considered using ERBs ?
(05:59:12 PM) jmelis: and later on, you can parse the xml in the deploy script:
(05:59:13 PM) jmelis: http://dev.opennebula.org/projects/opennebula/repository/revisions/master/entry/src/vmm_mad/remotes/kvm/deploy
(05:59:28 PM) jmelis: well, the core is written is c++, afaik there's no erb engine for c++
(05:59:47 PM) jmelis: but the option I'm describing shouldn't be hard at all to implement
(05:59:52 PM) vvuksan: i know
(06:00:05 PM) vvuksan: i am willing to modify LibVirtDriverKVM.cc
(06:00:09 PM) vvuksan: to add the proper field
(06:00:13 PM) vvuksan: would you commit if I do :-)
(06:00:34 PM) jmelis: if it's useful for everybody sure
(06:00:39 PM) jmelis: what are you trying to fix?
(06:00:51 PM) vvuksan: i had an issue with x86_64 VMs
(06:00:58 PM) vvuksan: since default is i686
(06:01:05 PM) jmelis: oh, sure,
(06:01:09 PM) jmelis: well you can modify this
(06:01:13 PM) vvuksan: it took me some time to figure out why instances weren't starting
(06:01:15 PM) vvuksan: i know
(06:01:18 PM) dtenney left the room (quit: Quit: dtenney).
(06:01:23 PM) vvuksan: having a console log is useful
(06:01:29 PM) vvuksan: openstack has one
(06:01:33 PM) dtenney [~dtenney@mkb044.apn.wireless-pennnet.upenn.edu] entered the room.
(06:01:38 PM) vvuksan: so you can at least figure out what happened
(06:01:48 PM) vvuksan: vs. editing deployment.0 adding vnc
(06:01:54 PM) vvuksan: then virsh destroy and virsh create
(06:02:45 PM) jmelis: here:
(06:02:46 PM) jmelis: http://dev.opennebula.org/projects/opennebula/repository/revisions/master/entry/src/vmm_mad/ssh/vmm_ssh_kvm.conf
(06:02:51 PM) vvuksan: i got that
(06:02:53 PM) vvuksan: :-)
(06:03:02 PM) vvuksan: i'm just saying
(06:03:09 PM) vvuksan: it would have been useful to have the console log
(06:03:19 PM) vvuksan: so that I can quickly check whether machine is booted or not
(06:03:22 PM) jmelis: to change the debug level you may want to change
(06:03:40 PM) jmelis: http://dev.opennebula.org/projects/opennebula/repository/revisions/master/entry/share/etc/defaultrc
(06:03:42 PM) jmelis: specifically
(06:03:48 PM) jmelis: ONE_MAD_DEBUG=1
(06:03:55 PM) jmelis: and the logs will be a lot more verbose
(06:04:04 PM) vvuksan: but it won't create the console log
(06:04:06 PM) vvuksan: for the VM
(06:04:46 PM) jfontan: the console is a bit hypervisor and configuration boundede
(06:04:55 PM) jmelis: how about using VNC?
(06:04:56 PM) jfontan: bounded
(06:05:07 PM) vvuksan: sure
(06:05:11 PM) vvuksan: you could do vnc as well
(06:05:16 PM) vvuksan: https://github.com/openstack/nova/blob/master/nova/virt/libvirt.xml.template#L102
(06:05:22 PM) vvuksan: this is what e.g. openstack does
(06:05:38 PM) jfontan: that can also be enabled
(06:05:49 PM) vvuksan: i think it's a reasonable default
(06:06:03 PM) jfontan: let me check for the exact place
(06:07:29 PM) jfontan: ok, the serial console is not directly supported by OPneNebula template
(06:07:37 PM) jfontan: but could be added using RAW parameters
(06:07:38 PM) vvuksan: you could stick it right here
(06:07:40 PM) vvuksan: http://dev.opennebula.org/projects/opennebula/repository/revisions/master/entry/src/vmm/LibVirtDriverKVM.cc#L608
(06:07:57 PM) jfontan: at the end of this file
(06:07:57 PM) jfontan: http://opennebula.org/documentation:rel2.2:template
(06:08:27 PM) vvuksan: hmm
(06:09:01 PM) vvuksan: k
(06:09:04 PM) vvuksan: i'll take a look
(06:09:06 PM) vvuksan: thx
(06:09:12 PM) jmelis: great
(06:09:17 PM) jfontan: but adding more console options are nice features, we have to check that
(06:09:28 PM) vvuksan: it's trivial :-)
(06:09:57 PM) jmelis: ok, let's move on to the next question
(06:10:11 PM) jmelis: koushik: How about providing autoscaling feature like similar to based on some load or trigger saving the VM to a file and launching multiple instances ( as requested by user ). Currently with save and resume option we can run only once.
(06:11:32 PM) jfontan: this is the job of a service manager
(06:11:42 PM) jfontan: OpenNebula sees VMs as black boxes
(06:12:12 PM) jfontan: and most of the time the values that will trigger a new VM deployment or shutting down an instance
(06:12:19 PM) jfontan: needs information that we cannot see
(06:12:36 PM) jfontan: that is, information that can only be read from the service itself
(06:13:15 PM) jfontan: for example for a HPC job queue the value to know when to start a new VM is the number of jobs in the queue
(06:13:21 PM) jfontan: or the waiting job in that queue
(06:13:25 PM) jfontan: not the CPU load
(06:13:49 PM) jfontan: as being 100% is the normal state of an HPC node
(06:14:15 PM) alzuin [bc5f4a48@gateway/web/freenode/ip.188.95.74.72] entered the room.
(06:14:20 PM) rsmontero left the room (quit: Read error: Operation timed out).
(06:14:27 PM) jfontan: so it should be another piece of software that should be getting this info from the service and issuing new VM's when needed
(06:14:34 PM) koushik: How abt Claudia serve the purpose of scaling the VM based on certain parameters
(06:14:53 PM) jfontan: Claudia job is exactly that
(06:15:13 PM) jfontan: you may take a look at http://stratuslab.org
(06:15:20 PM) koushik: thanks..
(06:15:31 PM) rsmontero [~ruben@pc-ruben.dacya.ucm.es] entered the room.
(06:16:02 PM) jfontan: both us and claudia people are involved in it
(06:16:13 PM) jfontan: and we are integrating both pieces of code
(06:16:23 PM) jfontan: any doubt with that?
(06:16:38 PM) koushik: its clear :)
(06:16:43 PM) jfontan: nice
(06:16:45 PM) jmelis: ok, next question:
(06:16:47 PM) jmelis: dtenney: Will version 2.4 work with VMware VCenter and ESXi 4.1?
(06:17:25 PM) tinova: Current version of VMware Addons (2.20) has been tested against ESXi 4.1, although not thoroughly
(06:17:52 PM) tinova: feedback on its compatibility (or lack thereof) are most welcome (the mailing list is a good candidate for this)
(06:18:14 PM) tinova: About vCenter integration, we will evaluate it certainly for the next Addon release
(06:18:19 PM) tinova: as well as vMotion support
(06:18:52 PM) dtenney: Great! will there be an vmware addon update with the new release?
(06:19:35 PM) tinova: we are not sure, but most likely
(06:20:13 PM) tinova: if it doesn't happen at the same time, it will most definitely happen a short time after
(06:20:40 PM) tinova: ok, any further doubts on this=?
(06:21:35 PM) dtenney: no, thank you.
(06:21:41 PM) jmelis: great
(06:21:45 PM) jmelis: next question:
(06:21:47 PM) jmelis: vbernetr: will there be an easy upgrade path from 2.2 to 2.4 ?
(06:22:19 PM) cmartin: yes, we are working on a new command line tool called onedb
(06:22:50 PM) mode (-o cmartin) by jmelis
(06:22:54 PM) cmartin: this tool will upgrade the DB schema and contents to the new version
(06:22:59 PM) mode (+o cmartin) by jmelis
(06:23:06 PM) cmartin: from a 2.0 or 2.2 installation
(06:23:21 PM) vbernetr: awesome :)
(06:23:47 PM) cmartin: any further comments about this tool?
(06:24:10 PM) vbernetr: nope :)
(06:24:31 PM) cmartin: btw, the issue is this one: http://dev.opennebula.org/issues/591
(06:24:43 PM) cmartin: you can follow the development there, or add any comments/requests
(06:25:01 PM) jmelis: ok, let's move to the next question:
(06:25:05 PM) jmelis: koushik: Does hybrid cloud setup required VPN between Opennebula and EC2. I am confused with two different documentation provided for hybrid cloud in opennebula. http://opennebula.org/documentation:rel2.2:ec2g this link refers to adding EC2 drivers to ONE makes hybrid config where as other doc contains info abt VPN and NIS required for hybrid ?
(06:25:59 PM) dmolina: yes, it does
(06:26:27 PM) dmolina: if you want to connect the ec2 vms with the one vms
(06:26:57 PM) dmolina: anyway you can define a new vnet in opennebual that contains the elastic ips to be used in the EC2
(06:27:22 PM) dmolina: therefore opennebula will handle the assignment of ips
(06:27:45 PM) koushik: Yeh I created a new vnet for ec2 with elastic IP and in opennebula shows ec2 as a host
(06:28:34 PM) koushik: I mean VPN is not mandatory for hybrid configuration ?
(06:29:09 PM) dmolina: you dont need a VPN for hybrid cloud
(06:29:28 PM) koushik: fine.. its clear
(06:29:39 PM) jmelis: ok, next question:
(06:29:43 PM) jmelis: koushik: I would like to demonstrate load balancing using hybrid cloud. I have opennebula private cloud setup working and have EC2 working account. I would like to know how I can acheive load balancing on VM using opennebula
(06:30:56 PM) jmelis: we don't quite get the question
(06:31:02 PM) jmelis: for instance
(06:31:07 PM) jmelis: if you're balancing a web server
(06:31:16 PM) jmelis: you can run your app in an VM in EC2
(06:31:25 PM) koushik: NGinx is mentioned for webservers. But if my VM contains other than webserver
(06:31:26 PM) jmelis: and use the VM's public ip address
(06:31:30 PM) jmelis: in your nginx balancer
(06:31:44 PM) jmelis: well that depends on your app
(06:31:52 PM) jmelis: what are you trying to load balance?
(06:32:04 PM) jmelis: it is a matter of configuration of the load balancer
(06:32:25 PM) jmelis: it doesn't really matter if the VM is running in a private cloud
(06:32:27 PM) jmelis: or in EC2
(06:32:34 PM) jmelis: as long as you have the service's ip address
(06:33:17 PM) koushik: Yeh . I am planning to distribute SMPP traffic between VM's
(06:33:59 PM) jmelis: unfortunately
(06:34:03 PM) koushik: NGINX can only distrbute HTTP messages so so I need to write my own balancer
(06:34:06 PM) jmelis: i'm not familiar with that protocol
(06:34:15 PM) jfontan: the thing is AFAIK there's not a silver bullet for load balancing
(06:34:25 PM) jfontan: it all depends on the server and protocols you are usings
(06:34:27 PM) jmelis: but what I'd do is find a SMPP balancer
(06:34:35 PM) jmelis: and configure it to use ec2 nodes
(06:34:41 PM) jfontan: and OpenNebula only knows aboyt VMs, IP's, images and so on
(06:35:09 PM) jfontan: the images with its services should be aware of these load balancing things and be configured acordingly
(06:35:39 PM) koushik: fine
(06:36:01 PM) jmelis: anyways
(06:36:11 PM) jmelis: opennebula doesn't know how to do load balancing
(06:36:14 PM) jmelis: it manages vms
(06:36:33 PM) jmelis: should we move on to the next question?
(06:36:42 PM) koushik: Sure
(06:36:55 PM) jmelis: vvuksan: how do I access network settings from the Vnet template inside the VM ie. looking at /mnt/context/context.sh I don't see any values for NETAMSK, GATEWAY
(06:37:22 PM) dmolina: you should add to you VM Template CONTEXT section
(06:37:26 PM) dmolina: the following lines:
(06:37:39 PM) dmolina: CONTEXT = [
(06:38:02 PM) dmolina: GATEWAY = $NETWORK[GATEWAY, NAME=\"Public\"]"
(06:38:04 PM) dmolina: ]
(06:38:28 PM) dmolina: you will be able to access this value inside the context.sh scipt
(06:38:42 PM) dmolina: http://opennebula.org/documentation:rel2.2:cong#defining_context
(06:39:04 PM) dmolina: but remember that first you must define this values in your vnet template
(06:40:30 PM) dmolina: is it clear?
(06:42:03 PM) jmelis: it seems vvuksan is not here
(06:42:09 PM) jmelis: let's move on to the next question:
(06:42:12 PM) jmelis: alzuin: I have a question too: will version 2.4 work with XEN 4.1? Expecially with the new xl command? Any plan to use remus for better VM HA?
(06:42:58 PM) dtenney left the room.
(06:43:04 PM) tinova: hi, i'm familiar with XCP, but not with XEN4.1
(06:43:11 PM) tinova: are they moving from "xm" to "xl" ?
(06:43:51 PM) alzuin: yes, but in this release there is also xm
(06:44:10 PM) alzuin: xl is the new command for future xen releases
(06:44:32 PM) tinova: we will certainly evaluate the change
(06:44:53 PM) tinova: it is not going to be a trivial task since other things are also changing
(06:45:12 PM) tinova: that is, is not as easy as changing xm request for xl requests, but is more going to be an integration task
(06:45:30 PM) tinova: bringing it closer to the interaction with vmware
(06:45:55 PM) tinova: ie, managing the software bridges, the storage resources, and so on
(06:46:30 PM) tinova: the answer I guess is that probably it won't happen in the next release (too close in time), although we are going to take steps to make the transition smooth
(06:47:15 PM) tinova: does this answer your question?
(06:48:24 PM) alzuin: You tell about XCP: any news for it?
(06:49:01 PM) jfontan: We have XCP in mind, as as you tell us about xen 4.1 they are going the same path
(06:49:27 PM) jfontan: but we still had no time to move in that direction
(06:49:43 PM) jfontan: but we will surely be supporting it in the future
(06:50:13 PM) alzuin: Ok, thanks.
(06:50:30 PM) jmelis: ok, moving on, last question:
(06:50:33 PM) jmelis: koushik: In hybrid environment Due to high load If I have to start a clone of existing running vm in ec2. I need to save a vm to disk and then convert it to AMI and then start it in EC2 using Open nebula commands(EC2 drivers). Is there any other method apart from this.
(06:50:52 PM) vbernetr left the room (quit: Ping timeout: 276 seconds).
(06:51:14 PM) dmolina: the new tool that have mentioned above
(06:51:33 PM) dmolina: will import ec2 amis to opennebula
(06:51:43 PM) dmolina: maybe we can implement the export functionality also
(06:51:51 PM) dmolina: but indeed the only thing you have to do
(06:52:04 PM) dmolina: is the bundle , upload ec2 sequence
(06:52:09 PM) dmolina: insisde the vm
(06:52:29 PM) tinova left the room (quit: Ping timeout: 240 seconds).
(06:53:16 PM) koushik: fine.. when is the probable time for release of import and export functionalities
(06:53:36 PM) dmolina: the import functionality will be included in the next release
(06:54:29 PM) koushik: fine