[Openstack] Java application over Openstack
Hi, Not sure whether this is the right place to post this question. I'm new to Openstack and trying to write a java application by using Openstack API. I read admin manual on the openstack site. It talks about using nova-manage to create the project, user and network. That's CLI interface. I'd like to do the same thing through openstack API. Is it doable? Is there any example available? I saw there are two api-server, one is nova-api, the other is nova-direct-api. Which one I should use? When I followed the manual, it seems that only nova-api is started. Should I start direct api server separately? Best, Rajesh ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Upload of Swift Glance in Debian Experimental
Soren Hansen wrote: 2011/6/2 Thomas Goirand tho...@goirand.fr: See above. It's opposite way. You're going to use the Ubuntu packaging proposed by Soren, and I'll derive from that in Debian. This sounds perfect! In fact, there's no policy in Ubuntu that mandates that ubuntu must be in the version string, even if the package isn't synced from Debian. If we were to drop that (and just use Debian-style versioning)) and you could commit directly to the packaging trunk, I think we could manage without any delta at all and you could simply sync the source package directly from Ubuntu. I think this would make for an excellent example of a different way of collaborating between Ubuntu and Debian. Indeed. I'm pretty sure Stefano would really like that example :) So all we need to make this happen is setting up an open-post (and therefore non-Launchpad) ML for the debian/ubuntu packagers group ? I'm pretty sure we can get something setup on lists.ubuntu.com, if everyone agrees. openstack-debian-packag...@lists.ubuntu.com ? -- Thierry Carrez (ttx) Release Manager, OpenStack ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Upload of Swift Glance in Debian Experimental
2011/6/6 Thierry Carrez thie...@openstack.org: Soren Hansen wrote: 2011/6/2 Thomas Goirand tho...@goirand.fr: So all we need to make this happen is setting up an open-post (and therefore non-Launchpad) ML for the debian/ubuntu packagers group ? I'm pretty sure we can get something setup on lists.ubuntu.com, if everyone agrees. openstack-debian-packag...@lists.ubuntu.com ? Why the debian- bit? It's meant to be shared, right? -- Soren Hansen | http://linux2go.dk/ Ubuntu Developer | http://www.ubuntu.com/ OpenStack Developer | http://www.openstack.org/ ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] Lunr question and GlusterFS integration
Hi All My research into the GlusterFS-OpenStack integration saga continues.. Going by this message: https://lists.launchpad.net/openstack/msg02667.html It is not very clear to me whether Lunr will be a replacement for Glance or will only be used for supporting application volumes. If I understand this correct, it is trying to overcome the iscsi-only access method currently provided for creating and accessing application volumes. Correct? With GlusterFS, my first milestone is to enable GlusterFS as an application volume backend, followed later by supporting GlusterFS as a VM store. So a deeper dive into Lunr would really help appreciate the integration points with Lunr. Are there any design docs available at this time? Thanks -Shehjar ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] Nova code coverage data
Hi, guys. It turns out that the coverage data we've seen on Jenkins has been somewhat optimistic. I thought Jenkins was configured to clean out the workspace in between runs, but it wasn't. This meant that the coverage data from previous runs wasn't deleted before running the test suite. So if line 1024 of file foo.py was covered in one revision and the next changed foo.py so that line 1024 wasn't covered anymore, it would still be recorded as covered. As you can see here: http://jenkins.openstack.org/job/nova-coverage/ the coverage percentage took quite a drop. Apologies for this. :( Thanks to Rohit Karajgi and Lorin Hochstein for putting me on to this. -- Soren Hansen | http://linux2go.dk/ Ubuntu Developer | http://www.ubuntu.com/ OpenStack Developer | http://www.openstack.org/ ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Integrating GlusterFS with OpenStack
My understanding is that the --instances_path will be the share to which the VM state will be synced on the source hypervisor. This synced image will then be used to restart the VM at the destination hypervisor. Ideally, I'd like to avoid having a local copy of the VM state in order to provide persistence in VMs so that if the hypervisor fails or crashes, etc. the VM state on a Gluster volume can be used to restart it. I am not sure whether this is possible in OpenStack yet or if KVM will even support using a shared mount as a VM state storage. This is something I will be looking into. Of course, some references to docs, emails will be nice . Thanks. From: masumo...@nttdata.co.jp [masumo...@nttdata.co.jp] Sent: Monday, June 06, 2011 6:23 AM To: Shehjar Tikoo Cc: openstack@lists.launchpad.net; so...@linux2go.dk Subject: RE: [Openstack] Integrating GlusterFS with OpenStack Hi, When doing live migration, we could let KVM migrate the VM state and somehow tell gluster to migrate the disk image to the new host (to maintain locality between the virtual machine and its storage). Gluster already has the means to do live rebalancing, but I don't believe an API to specify I'd like for this particular file to reside on this particular brick. Please make it so. That would *rock*. In current openstack live migration feature, a flag --instances_path specifies the path which has to be on shared storage. Live re-balancing sounds like good idea, and I think at least, just --instances_path is on the glusterFS is good option to users. If you want any help, my colleague and me would love to do ^^; -Original Message- From: openstack-bounces+masumotok=nttdata.co...@lists.launchpad.net [mailto:openstack-bounces+masumotok=nttdata.co...@lists.launchpad.ne t] On Behalf Of Soren Hansen Sent: Monday, June 06, 2011 7:40 AM To: Shehjar Tikoo Cc: openstack@lists.launchpad.net Subject: Re: [Openstack] Integrating GlusterFS with OpenStack 2011/6/3 Shehjar Tikoo shehj...@gluster.com: We're aiming to integrate Gluster with Openstack in a way that allows GlusterFS to be used as the storage for application volumes as well as VMs. I am interested in hearing your ideas on how such an integration can be performed. Being the openstack experts, could you please share some issues I should be considering, the problems I may run into or anything else you think I should know. One thing I'd love to be able to do is store my virtual disks on a gluster filesystem that is shared across my compute nodes. The data would be stored on the host that is actually running the virtual machine (think NUFA scheduler). When doing live migration, we could let KVM migrate the VM state and somehow tell gluster to migrate the disk image to the new host (to maintain locality between the virtual machine and its storage). Gluster already has the means to do live rebalancing, but I don't believe an API to specify I'd like for this particular file to reside on this particular brick. Please make it so. That would *rock*. -- Soren Hansen| http://linux2go.dk/ Ubuntu Developer| http://www.ubuntu.com/ OpenStack Developer | http://www.openstack.org/ ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Lunr question and GlusterFS integration
On Mon, 6 Jun 2011 09:59:19 + Shehjar Tikoo shehj...@gluster.com wrote: It is not very clear to me whether Lunr will be a replacement for Glance I don't think so. or will only be used for supporting application volumes. I'm not sure the meaning of application volumes but it's a replacement of the current nova-volume, provinding volume service for VMs. If I understand this correct, it is trying to overcome the iscsi-only access method currently provided for creating and accessing application volumes. Correct? nova-volume already supports non iSCSI protocols, AOE, Sheepdog, Ceph RBD. I think that you start working on GlusterFS integration with nova-volume right now. You can work with Lunr later (when it's released). ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] IPMI plugin for Scheduler
I am planning to develop an IPMI power control plugin for Nova Scheduler and would like to solicit feature requests and implementation suggestions from the list. This is still in early planning stage so any and all feedback is welcome. Thanks, Kyle Dunn USC-ISI Intern ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] IPMI plugin for Scheduler
Kyle, Please wait until sprint planning meeting since we don't know if this user story will make it in this sprint. We can start working if the user story is included in this (or future) sprint task list. Thanks, Joseph - Original Message - From: Kyle Dunn kd...@isi.edu To: openstack@lists.launchpad.net Sent: Monday, June 6, 2011 7:45:51 AM Subject: [Openstack] IPMI plugin for Scheduler I am planning to develop an IPMI power control plugin for Nova Scheduler and would like to solicit feature requests and implementation suggestions from the list. This is still in early planning stage so any and all feedback is welcome. Thanks, Kyle Dunn USC-ISI Intern ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] Jenkins - nova-pylint-errors
I've been working on cleaning up pylint errors and have a concern regarding: E1101: %s %r has no %r member Used when a variable is accessed for an unexistent member. 97 (59%) of the 169 errors in the nova pylint errors job on jenkins are of the type E1101. 82/97 are located under the sqlalchemy directory. This is (mostly) because pylint doesn't understand most of the ways classes and functions in the database code are getting members even though they are by no means incorrect. One example that pylint complains about is this: nova/db/sqlalchemy/api.py Class 'Network' has no 'fixed_ips' member 1322@require_admin_context 1323def network_get_by_instance(_context, instance_id): 1324session = get_session() 1325rv = session.query(models.Network).\ 1326 filter_by(deleted=False).\ ! 1327 join(models.Network.fixed_ips).\ 1328 filter_by(instance_id=instance_id).\ 1329 filter_by(deleted=False).\ 1330 first() But Network does have the attribute fixed_ips because it is created here: nova/db/sqlalchemy/models.py 536: network = relationship(Network, backref=backref('fixed_ips')) What do you guys think about setting jenkins to ignore E1101 or at least ignore it for nova/db/sqlalchemy? If we ever want to get our pylint errors down to zero so we can start enforcing it, we need to either do something like above or insert pylint ignore statements throughout all the code where the error is not really an error. ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Jenkins - nova-pylint-errors
Hey Vish, I may be missing something but when I test with pylint locally, I am using the pylintrc located in trunk. The only cleanup I am doing is derived directly from looking at the nova-pylint-errors job on Jenkins. Alex -Original Message- From: Vishvananda Ishaya vishvana...@gmail.com Sent: Monday, June 6, 2011 11:42am To: Alex Meade alex.me...@rackspace.com Cc: openstack@lists.launchpad.net Subject: Re: [Openstack] Jenkins - nova-pylint-errors Hey alex, Justin came up with a pretty reasonable set of defaults for pylint that tends to only show real error messages. It is the nova-pylint-errors job in jenkins. It looks like this error has already been removed from the list (although there is another one there about medatada #690) Perhaps you should start your cleanup by fixing the ones still showing up in that job rather than the default pylint output? Vish On Jun 6, 2011, at 6:08 AM, Alex Meade wrote: I've been working on cleaning up pylint errors and have a concern regarding: E1101: %s %r has no %r member Used when a variable is accessed for an unexistent member. 97 (59%) of the 169 errors in the nova pylint errors job on jenkins are of the type E1101. 82/97 are located under the sqlalchemy directory. This is (mostly) because pylint doesn't understand most of the ways classes and functions in the database code are getting members even though they are by no means incorrect. One example that pylint complains about is this: nova/db/sqlalchemy/api.py Class 'Network' has no 'fixed_ips' member 1322@require_admin_context 1323def network_get_by_instance(_context, instance_id): 1324session = get_session() 1325rv = session.query(models.Network).\ 1326 filter_by(deleted=False).\ ! 1327 join(models.Network.fixed_ips).\ 1328 filter_by(instance_id=instance_id).\ 1329 filter_by(deleted=False).\ 1330 first() But Network does have the attribute fixed_ips because it is created here: nova/db/sqlalchemy/models.py 536: network = relationship(Network, backref=backref('fixed_ips')) What do you guys think about setting jenkins to ignore E1101 or at least ignore it for nova/db/sqlalchemy? If we ever want to get our pylint errors down to zero so we can start enforcing it, we need to either do something like above or insert pylint ignore statements throughout all the code where the error is not really an error. ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
[Openstack] doc/web meeting, 0200 UTC
Hi all - If you're interested in documentation and the web content for OpenStack, we're having our first monthly meeting tonight in #openstack-meeting on IRC. The pattern is the first Monday of the month at 0200 UTC. Please feel free to add items to the agenda at http://wiki.openstack.org/Meetings/DocTeamMeeting. I'll report back to the developer meeting after we hold our meeting. Thanks, Anne *Anne Gentle* a...@openstack.org my blog http://justwriteclick.com/ | my bookhttp://xmlpress.net/publications/conversation-community/| LinkedIn http://www.linkedin.com/in/annegentle | Delicioushttp://del.icio.us/annegentle| Twitter http://twitter.com/annegentle ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Integrating GlusterFS with OpenStack
gfs2/rgmanager works reliably for kvm so i don't see why gluster wouldn't. On Mon, Jun 6, 2011 at 5:20 AM, Shehjar Tikoo shehj...@gluster.com wrote: My understanding is that the --instances_path will be the share to which the VM state will be synced on the source hypervisor. This synced image will then be used to restart the VM at the destination hypervisor. Ideally, I'd like to avoid having a local copy of the VM state in order to provide persistence in VMs so that if the hypervisor fails or crashes, etc. the VM state on a Gluster volume can be used to restart it. I am not sure whether this is possible in OpenStack yet or if KVM will even support using a shared mount as a VM state storage. This is something I will be looking into. Of course, some references to docs, emails will be nice . Thanks. From: masumo...@nttdata.co.jp [masumo...@nttdata.co.jp] Sent: Monday, June 06, 2011 6:23 AM To: Shehjar Tikoo Cc: openstack@lists.launchpad.net; so...@linux2go.dk Subject: RE: [Openstack] Integrating GlusterFS with OpenStack Hi, When doing live migration, we could let KVM migrate the VM state and somehow tell gluster to migrate the disk image to the new host (to maintain locality between the virtual machine and its storage). Gluster already has the means to do live rebalancing, but I don't believe an API to specify I'd like for this particular file to reside on this particular brick. Please make it so. That would *rock*. In current openstack live migration feature, a flag --instances_path specifies the path which has to be on shared storage. Live re-balancing sounds like good idea, and I think at least, just --instances_path is on the glusterFS is good option to users. If you want any help, my colleague and me would love to do ^^; -Original Message- From: openstack-bounces+masumotok=nttdata.co...@lists.launchpad.net [mailto:openstack-bounces+masumotok=nttdata.co...@lists.launchpad.ne t] On Behalf Of Soren Hansen Sent: Monday, June 06, 2011 7:40 AM To: Shehjar Tikoo Cc: openstack@lists.launchpad.net Subject: Re: [Openstack] Integrating GlusterFS with OpenStack 2011/6/3 Shehjar Tikoo shehj...@gluster.com: We're aiming to integrate Gluster with Openstack in a way that allows GlusterFS to be used as the storage for application volumes as well as VMs. I am interested in hearing your ideas on how such an integration can be performed. Being the openstack experts, could you please share some issues I should be considering, the problems I may run into or anything else you think I should know. One thing I'd love to be able to do is store my virtual disks on a gluster filesystem that is shared across my compute nodes. The data would be stored on the host that is actually running the virtual machine (think NUFA scheduler). When doing live migration, we could let KVM migrate the VM state and somehow tell gluster to migrate the disk image to the new host (to maintain locality between the virtual machine and its storage). Gluster already has the means to do live rebalancing, but I don't believe an API to specify I'd like for this particular file to reside on this particular brick. Please make it so. That would *rock*. -- Soren Hansen| http://linux2go.dk/ Ubuntu Developer| http://www.ubuntu.com/ OpenStack Developer | http://www.openstack.org/ ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Java application over Openstack
On 06/06/11 02:30, Rajesh Mohan wrote: Hi, Not sure whether this is the right place to post this question. I'm new to Openstack and trying to write a java application by using Openstack API. I read admin manual on the openstack site. It talks about using nova-manage to create the project, user and network. That's CLI interface. I'd like to do the same thing through openstack API. Is it doable? Is there any example available? I saw there are two api-server, one is nova-api, the other is nova-direct-api. Which one I should use? When I followed the manual, it seems that only nova-api is started. Should I start direct api server separately? Best, Rajesh ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp You can find API information in OpenStack DeveloperGuide API at http://docs.openstack.org/. nova-direct-api I think (I'm not really sure) is used only with nova command line client, so you don't need it. Bye ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Java application over Openstack
Thanks Rafael and Jay. By following the openstack installation manual, nova-direct-api is not installed. To install it, should I just download openstack package and copy the nova-direct-api and stack to the /usr/bin/ directory? The nova-api seems have limited function to call, but nova-direct-api seems can call any function, right? Best, Rajesh On Jun 6, 2011 12:00pm, Jay Pipes jaypi...@gmail.com wrote: 2011/6/6 Rafael Durán Castañeda rafadurancastan...@gmail.com: You can find API information in OpenStack DeveloperGuide API at http://docs.openstack.org/. nova-direct-api I think (I'm not really sure) is used only with nova command line client, so you don't need it. Hi Rafael, Actually, the nova-direct-api is a server that publishes an alternate HTTP interface called the Direct API. This API can be accessed with the bin/stack CLI tool Cheers, jay ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Java application over Openstack
On Mon, Jun 6, 2011 at 3:15 PM, mohan.rajes...@gmail.com wrote: Thanks Rafael and Jay. By following the openstack installation manual, nova-direct-api is not installed. To install it, should I just download openstack package and copy the nova-direct-api and stack to the /usr/bin/ directory? Hmm, that's interesting. Termie, is there a reason we didn't want nova-direct-api installed where the other bins are installed? Was the intention to keep nova-direct-api as more of a developer-focused thing? The nova-api seems have limited function to call, but nova-direct-api seems can call any function, right? nova-api publishes the EC2 and official OpenStack HTTP APIs. nova-direct-api publishes an alternate HTTP API that merely exposes the actual methods of the various controllers as URL endpoints. It's just a different way of accessing the services... -jay ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp
Re: [Openstack] Lunr question and GlusterFS integration
On Jun 6, 2011, at 3:17 AM, FUJITA Tomonori wrote: On Mon, 6 Jun 2011 09:59:19 + Shehjar Tikoo shehj...@gluster.com wrote: It is not very clear to me whether Lunr will be a replacement for Glance I don't think so. Glance is the image store which uses s3/swift api's to store machine images. Lunr is an effort to take the block volume code that is currently in nova and separate it into it's own more generic service. or will only be used for supporting application volumes. I'm not sure the meaning of application volumes but it's a replacement of the current nova-volume, provinding volume service for VMs. Yep, exactly. I believe Shehjart was referring to EBS style block storage, which nova-volume currently provides (and Lunr will replace in the future most likely). If I understand this correct, it is trying to overcome the iscsi-only access method currently provided for creating and accessing application volumes. Correct? nova-volume already supports non iSCSI protocols, AOE, Sheepdog, Ceph RBD. I think that you start working on GlusterFS integration with nova-volume right now. You can work with Lunr later (when it's released). Agreed - it seems most reasonable to begin this work in nova-volume and then add support to Lunr when it is closer to being ready for prime time. ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp ___ Mailing list: https://launchpad.net/~openstack Post to : openstack@lists.launchpad.net Unsubscribe : https://launchpad.net/~openstack More help : https://help.launchpad.net/ListHelp