[openstack-dev] [Horizon] The source language of Horizon in Transifex
Hi, Now the source language of Horizon in Transifex is set as en_US, not en. So when pulling the translations from Transifex, there will be some dummy characters in en.po, which will cause errors in unit tests. I don't find a way to change the setting in Transifex. I think the only way to fix it is to re-upload the resources with source language setting as en, and delete the existing resources. Please let me know if Horizon development team know the issue and have any plans to fix it. Thanks Regards Daisy___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Proposal for approving Auto HA development blueprint.
Hi yongiman: I wander what’s the difference between your ‘auto HA’ API and ‘evacuate’ Lingxian Kong Huawei Technologies Co.,LTD. IT Product Line CloudOS PDU China, Xi'an Mobile: +86-18602962792 Email: konglingx...@huawei.com From: yongi...@gmail.com [mailto:yongi...@gmail.com] Sent: Tuesday, August 13, 2013 9:12 PM To: OpenStack Development Mailing List Cc: OpenStack Development Mailing List Subject: Re: [openstack-dev] 答复: Proposal for approving Auto HA development blueprint. For realizing auto HA function, we need monitoring service like ceilometer. Ceilometer monitors status of compute nodes ( network interface..connection, healthcheck,,etc,,) What I focus on is that this operation goes on automatically. Nova expose auto ha API. When nova received a auto api call. VMs automatically migrate to auto ha host.( which is extra compute node for only auto ha) All of information of auto ha is stored in auto_ha_hosts table. In this tables, used column of auto ha hosts is changed to true Administrator check broken compute node and fix( or replace ) the compute node. After fixing the compute node, VMs is migrating to operating compute nodes. Now auto ha host is empty again. When the number of runnning VMs in the auto ha host is zero, used column is replaced to false for using again by periodic task. Combination with monitoring service is important. Howerver in this blueprint, I want to realize nova's auto ha operation. My wiki page is still building now, I will fill out as soon as possbile. I am expecting your advices . Thank you very much~! Sent from my iPad On 2013. 8. 13., at 오후 8:01, balaji patnala patnala...@gmail.commailto:patnala...@gmail.com wrote: Potential candidate as new service like Ceilometer, Heat etc for OpenStack and provide High Availability of VMs. Good topic to discuss at Summit for implementation post Havana Release. On Tue, Aug 13, 2013 at 12:03 PM, Alex Glikson glik...@il.ibm.commailto:glik...@il.ibm.com wrote: Agree. Some enhancements to Nova might be still required (e.g., to handle resource reservations, so that there is enough capacity), but the end-to-end framework probably should be outside of existing services, probably talking to Nova, Ceilometer and potentially other components (maybe Cinder, Neutron, Ironic), and 'orchestrating' failure detection, fencing and recovery. Probably worth a discussion at the upcoming summit. Regards, Alex From:Konglingxian konglingx...@huawei.commailto:konglingx...@huawei.com To:OpenStack Development Mailing List openstack-dev@lists.openstack.orgmailto:openstack-dev@lists.openstack.org, Date:13/08/2013 07:07 AM Subject:[openstack-dev] 答复: Proposal for approving Auto HA development blueprint. Hi yongiman: Your idea is good, but I think the auto HA operation is not OpenStack’s business. IMO, Ceilometer offers ‘monitoring’, Nova offers ‘evacuation’, and you can combine them to realize HA operation. So, I’m afraid I can’t understand the specific implementation details very well. Any different opinions? 发件人: yongi...@gmail.commailto:yongi...@gmail.com [mailto:yongi...@gmail.com] 发送时间: 2013年8月12日 20:52 收件人: openstack-dev@lists.openstack.orgmailto:openstack-dev@lists.openstack.org 主题: Re: [openstack-dev] Proposal for approving Auto HA development blueprint. Hi, Now, I am developing auto ha operation for vm high availability. This function is all progress automatically. It needs other service like ceilometer. ceilometer monitors compute nodes. When ceilometer detects broken compute node, it send a api call to Nova, nova exposes for auto ha API. When received auto ha call, nova progress auto ha operation. All auto ha enabled VM where are running on broken host are all migrated to auto ha Host which is extra compute node for using only Auto-HA function. Below is my blueprint and wiki page. Wiki page is not yet completed. Now I am adding lots of information for this function. Thanks https://blueprints.launchpad.net/nova/+spec/vm-auto-ha-when-host-broken https://wiki.openstack.org/wiki/Autoha___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.orgmailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org
Re: [openstack-dev] [Horizon] The source language of Horizon in Transifex
Hi Daisy, Ying Chun Guo guoyi...@cn.ibm.com wrote: Now the source language of Horizon in Transifex is set as en_US, not en. So when pulling the translations from Transifex, there will be some dummy characters in en.po, which will cause errors in unit tests. I don't find a way to change the setting in Transifex. I think the only way to fix it is to re-upload the resources with source language setting as en, and delete the existing resources. Please let me know if Horizon development team know the issue and have any plans to fix it. Thanks We have a bug open for tracking this, 1179526 [0]. Akihiro Motoki suggested a temporary change that prevents the issues we saw with unit tests, which we should implement. I think at this point we were planning on reaching out to our contacts at Transifex (...if we have any?) to see if it would be possible to fix this without having to delete and recreate the project. Regards, Julie [0] https://bugs.launchpad.net/horizon/+bug/1179526 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] Question about get_meters query using a JOIN
On Tue, Aug 13 2013, Thomas Maddox wrote: I was curious about why we went for a JOIN here rather than just using the meter table initially? https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/impl_sqlalchemy.py#L336-L391. Doug had mentioned that some performance testing had gone on with some of these queries, so before writing up requests to change this to the meter table only, I wanted to see if this was a result of that performance testing? Like the JOIN was less expensive than a DISTINCT. Because the SQL driver has been historically a straight conversion from MongoDB, who was working that way. The MongoDB driver had these 2 collections, meter and resources, and used both of them to construct these data. The original idea was to register resources in the eponym table, and use it to filter. Unfortunately, that didn't work out, especially with APIv2 that allows to filter on many things. Among other things, there was timestamp fields in the resource table that were used to filter based on timestamp, but this was a naive implementation that failed pretty soon, and we removed and replace with a more solid one based on the meter table. What you see here are the rest of that time, and in the end, this resource table should be removed AFAICT, in MongoDB also. -- Julien Danjou ;; Free Software hacker ; freelance consultant ;; http://julien.danjou.info signature.asc Description: PGP signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Proposal for approving Starting by scheduler development blueprint.
I’m interested in this, want to hear others’ opinions. Lingxian Kong Huawei Technologies Co.,LTD. IT Product Line CloudOS PDU China, Xi'an Mobile: +86-18602962792 Email: konglingx...@huawei.com From: cosmos cosmos [mailto:cosmos0...@gmail.com] Sent: Wednesday, August 14, 2013 10:32 AM To: openstack-dev@lists.openstack.org Subject: [openstack-dev] Proposal for approving Starting by scheduler development blueprint. Hello. My name is Rucia for Samsung SDS. Now, I am developing Start Logic by nova-scheduler for efficient resources of host. This function is already implemented in folsom release version. It is used for the iscsi target such as HP san storage. This is slightly different from the original version. If you start the instance after stop, the instance will started at optimal Compute host. The selected host is through the nova-scheduler. Before Logic 1. Do not use the scheduler originally in start logic of Openstack Nova 2. Start on the host where the instance is created After Logic 1. When the stopped instance start, Changed to start from the hosts that is selected by nova-scheduler 2. When the VM starts, Check the resources through check_resource_limit() Pros - You can use resources efficiently - When you start a virtual machine, You can solve the problem that is error caused by the lack of resources on a host. Below is my blueprint and wiki page. Thanks https://blueprints.launchpad.net/nova/+spec/start-instance-by-scheduler https://wiki.openstack.org/wiki/Start ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Difference between RBAC polices thats stored in policy.json and policies that can be created using openstack/identity/v3/policies
Hi Sudheesh, Using v3/policies is just a way of allowing other keystone projects (nova, glance) etc. a place to centrally store/access their policy files. Keystone does not interpret any of the data you store here - it is simply acting as a central repository (where you can store a big blob of data that is, in effect, your policy file). So the only place you can set policies is in the policy file. Henry On 13 Aug 2013, at 08:22, sudheesh sk wrote: Hi , I am trying to understand Difference between RBAC polices thats stored in policy.json and policies that can be created using openstack/identity/v3/policies. I got answer from openstack forum that I can use both DB and policy.json based implementation for RBAC policy management. Can you please tell me how to use DB based RBAC ? I can elaborate my question 1. In policy.json(keystone) I am able to define rule called - admin_required 2. Similarly I can define rules line custome_role_required 3. Then I can add this rule against each services (like for eg : identity:list_users = custom_role_required How can I use this for DB based RBAC policies? Also there are code like self.policy_api.enforce(context, creds, 'admin_required', {}) in many places (this is in wsgi.py) How can I utilize the same code and at the same time move the policy definition to DB Thanks a million, Sudheesh ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] Concerning get_resources/get_meters and the Ceilometer API
On Tue, Aug 13 2013, Thomas Maddox wrote: Hi Thomas, * Driving get_resources() with the Meter table instead of the Resource table. This is mainly because of the additional filtering available in the Meter table, which allows us to satisfy a use case like getting a list of resources a user had during a period of time to get meters to compute billing with. The semantics are tripping me up a bit; the question this boiled down to for me was: why use a resource query to get meters to show usage by a tenant? I was curious about why we needed the timestamp filtering when looking at Resources, and why we would use Resource as a way to get at metering data, rather than a Meter request itself? This was answered by resources being the current vector to get at metering data for a tenant in terms of resources, if I understood correctly. * With this implementation, we have to do aggregation to get at the discrete Resources (via the Meter table) rather than just filtering the already distinct ones in the Resource table. I think I already answered that in a previous email when I said drop the resource table. :) * This brought up some confusion with the API for me with the major use cases I can think of: * As a new consumer of this API, I would think that /resource/resource_id would get me details for a resource, e.g. current state, when it was created, last updated/used timestamp, who owns it; not the attributes from the first sample to come through about it s/first/last/ actually I wouldn't disagree if you had such improvements to propose. However, we're pretty flexible in Ceilometer as we don't allow only metering OpenStack; be careful of somethings like who owns it might change in other systems than OpenStack for example, depending on the time range you filter on. * I would think that /meter/?q.field=resource_idq.value=resource_id ought to get me a list of meter(s) details for a specific resource, e.g. name, unit, and origin; but not a huge mixture of samples. That could be a nice improvement indeed. * Additionally /meter/?q.field=user_idq.value=user_id would get me a list of all meters that are currently related to the user Same as above. * The ultimate use case, for billing queries, I would think that /meter/meter_id/statistics?time filtersuser(resource_id) would get me the measurements for that meter to bill for. We'd like that too, though it's not always perfect since we don't handle the different counter types explicitly. If I understand correctly, one main intent driving this is wanting to avoid end users having to write a bunch of API requests themselves from the billing side and instead just drill down from payloads for each resource to get the billing information for their customers. It also looks like there's a BP to add grouping functionality to statistics queries to allow us this functionality easily (this one, I think: https://blueprints.launchpad.net/ceilometer/+spec/api-group-by). Yes. I'm new to this project, so I'm trying to get a handle on how we got here and maybe offer some outside perspective, if it's needed or wanted. =] I think you got the picture right. We're trying to improve the API, but we always happy to get help. There's a sort of meta blueprint: https://blueprints.launchpad.net/ceilometer/+spec/api-v2-improvement With various ideas to improve the API. It's assigned to me, though I didn't implement most of the ideas there, and won't probably have time to implement them all, so feel free to contribute! -- Julien Danjou /* Free Software hacker * freelance consultant http://julien.danjou.info */ signature.asc Description: PGP signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [ANVIL] Missing openvswitch dependency for basic-neutron.yaml persona
Thanks for the links, pretty useful. I do understand the process, but I have to admin I don't catch which Cheetah placeholder I would use for doing a big 'if' statement conditioning the package openstack-neutron-openvswitch on the core_plugin yaml option. As you said, this is not enough, if asked, openvswitch should also either be compiled or fetched from RDO. I filed a bug : https://bugs.launchpad.net/anvil/+bug/1212165 Anyway, I'm pretty much interested in doing the 1. you mentioned, I still need to understand things, tho. Could you be more precise on the way the spec files are populated ? Thanks, -Sylvain Le 13/08/2013 19:55, Joshua Harlow a écrit : Haha, no problem. Darn time differences. So some other useful links that I think will be helpful. - https://github.com/stackforge/anvil/blob/master/conf/templates/packaging/specs/openstack-neutron.spec This one is likely the biggest part of the issue, since it is the combination of all of neutron into 1 package (which has sub-packages). - One of those sub-packages is https://github.com/stackforge/anvil/blob/master/conf/templates/packaging/specs/openstack-neutron.spec#L274 This is pulling in the openvswitch part, that I think u don't want (at least not always want, it's wanted if neutron is going to use it, which under certain plugins it will). As you've seen it likely shouldn't be installing/needing that if https://github.com/stackforge/anvil/blob/master/anvil/components/configurators/neutron_plugins/linuxbridge.py is used. This should be coming from the following config (which will come from the yaml files) 'get_option' 'call': https://github.com/stackforge/anvil/blob/master/anvil/components/configurators/neutron.py#L49 So I think what can be done is a couple of things: 1. Don't include sub-packages that we don't want (the spec files are cheetah http://www.cheetahtemplate.org/ templates, so this can be done dynamically). 2. See if there is a way to make yum (or via yyoom) not pull in the dependencies for a sub-package when it won't be used (?) 3. Always build openvswitch (not as preferable) and include it (https://github.com/stackforge/anvil/blob/master/tools/build-openvswitch.sh) * I think the RDO repos might have some of these components. * http://openstack.redhat.com/Frequently_Asked_Questions#For_which_distributions_does_RDO_provide_packages.3F * This means we can just include the RDO repo rpm (like epel and use that openvswitch version there) instead of build your own. Hope some of this offers some good pointers. -Josh From: Sylvain Bauza sylvain.ba...@bull.net mailto:sylvain.ba...@bull.net Date: Tuesday, August 13, 2013 9:52 AM To: Joshua Harlow harlo...@yahoo-inc.com mailto:harlo...@yahoo-inc.com Cc: OpenStack Development Mailing List openstack-dev@lists.openstack.org mailto:openstack-dev@lists.openstack.org Subject: Re: [openstack-dev] [ANVIL] Missing openvswitch dependency for basic-neutron.yaml persona Well, then I have to read thru the docs to see how it can be done thru a config option... =) Nope, I won't be able to catch you up on IRC, time difference you know :-) Anyway, let me go thru it, I'll try to sort it out. I RTFM'd all the anvil docs, but do you have any other pointer for me ? Thanks, -Sylvain Le 13/08/2013 18:39, Joshua Harlow a écrit : Well open switch is likely needed still when it's really needed right? So I think there is a need for it. It just might have to be a dynamic choice (based on a config option) instead of a static choice. Make sense?? The other personas don't use neutron so I think that's how they work, since nova-network base functionality still exists. Any patches would be great, will be on irc soon if u want to discuss more. Josh Sent from my really tiny device... On Aug 13, 2013, at 9:23 AM, Sylvain Bauza sylvain.ba...@bull.net mailto:sylvain.ba...@bull.net wrote: Do you confirm the basic idea would be to get rid of any openvswitch reference in rhel.yaml ? If so, wouldn't it be breaking other personas ? I can provide a patch so the team would review it. -Sylvain Le 13/08/2013 17:57, Joshua Harlow a écrit : It likely shouldn't be needed :) I haven't personally messes around with the neutron persona to much and I know that it just underwent the great rename of 2013 so u might be hitting issues due to that. Try seeing if u can adjust the yaml file and if not I am on irc to help more. Sent from my really tiny device... On Aug 12, 2013, at 9:14 AM, Sylvain Bauza sylvain.ba...@bull.net mailto:sylvain.ba...@bull.net wrote: Hi, ./smithy -a install -p conf/personas/in-a-box/basic-neutron.yaml is failing because of openvswitch missing. See logs here [1]. Does anyone knows why openvswitch is needed when asking for linuxbridge in components/neutron.yaml ? Shall I update distros/rhel.yaml ? -Sylvain [1] : http://pastebin.com/TFkDrrDc
Re: [openstack-dev] Swift, netifaces, PyPy, and cffi
On Tue, Aug 13, 2013 at 11:58 PM, Alex Gaynor alex.gay...@gmail.com wrote: a) cffi-only approach, this is obviously the simplest approach, and works everywhere (assuming you can install a PPA, use pip, or similar for cffi) b) wait until the next LTS to move to this approach (requires waiting until 2014 for PyPy support) c) Support using either netifaces or cffi: most complex, and most code, plus one or the other dependencies aren't well supported by most tools as far as I know. I think it sucks for large public cloud already in production to have to package a new python module and having to deploy it when most of them are pretty conservative with what goes in production, and we can't really assume they are going to update all their LTS to the new version when it's going to be released. Having said that there is people interested to know what performances improvement can come out from your work (thanks for that) to get PyPy support If there is a vote I would go for c) then, potentially in the future we could unify it to just use cffi. When you are saying 'plus one or the other dependencies aren't well supported by most tools as far as I know' are you talking about apt/pip etc..? I would assume that this is up to the distro packager which knows that the distro targetted have the cffi package. A far goes for pip-requires I would conservatively keep the dependence to netifaces. Chmouel. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] GPU passthrough support blueprints for OpenStack
Hi Brian, Instead of specific GPU pass-through there are blueprints for generic PCI pass through at https://blueprints.launchpad.net/nova/+spec/pci-passthrough-base - these are being worked on for Havana and we're hopeful they will get in (possibly libvirt only) In terms of vGPU rather than whole-GPU pass through, the slides are correct that XenServer support is very nearly here: http://www.xenserver.org/discuss-virtualization/q-and-a/nvidia-vgpu-on-xenserver-eg-gpu-hypervisor.html Support for static allocation of vGPU is likely going to be the first phase - and adding it as a resource similar to RAM/CPU isn't likely to be proposed until a second phase. As such, I think the integration should be very similar to PCI pass through above. Bob -Original Message- From: Brian Schott [mailto:brian.sch...@nimbisservices.com] Sent: 13 August 2013 23:07 To: openstack-dev@lists.openstack.org Subject: [openstack-dev] GPU passthrough support blueprints for OpenStack Are there more recent blueprints related to adding GPU pass-through support? All that I can find are some stale blueprints that I created around the Cactus timeframe (while wearing a different hat) that are pretty out of date. I just heard a rumor that folks are doing Nvidia GRID K2 GPU passthrough with KVM successfully using linux 3.10.6 kernel with RHEL. In addition, Lorin and I did some GPU passthrough testing back in the spring with GRID K2 on HyperV, libvirt+xen, and XenServer. Slides are here: http://www.slideshare.net/bfschott/nimbis-schott- openstackgpustatus20130618 The virtualization support for GPU-enabled virtual desktops and GPGPU seems to have stabilized this year for server deployments. How is this going to be supported in OpenStack? Brian - Brian Schott, CTO Nimbis Services, Inc. brian.sch...@nimbisservices.com ph: 443-274-6064 fx: 443-274-6060 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [Neutron] Ceilometer support
[cid:image001.jpg@01CE98FD.2A08F120] inline: image001.jpg___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] Does Nova v2API need to know about any Neutron objects apart from ports ? (was RE: [nova] security_groups extension in nova api v3)
On Aug 13, 2013, at 2:11 AM, Day, Phil wrote: If we really want to get clean separation between Nova and Neutron in the V3 API should we consider making the Nov aV3 API only accept lists o port ids in the server create command ? That way there would be no need to every pass security group information into Nova. Any cross project co-ordination (for example automatically creating ports) could be handled in the client layer, rather than inside Nova. Server create is always (until there's a separate layer) going to go cross project calling other apis like neutron and cinder while an instance is being provisioned. For that reason, I tend to think it's ok to give some extra convenience of automatically creating ports if needed, and being able to specify security groups. I think there's a difference between the current interaction with Cinder, where Nova will accept a UUID of a cinder object to attach to, and the interaction with Neutron where Nova actually creates the Port Objects in Neutron, and as a result also needs to accept parameters that are attributes of the foreign object. Fundamentally the relationship between objects in Nova and Neutron is the instanceport mapping - so I'm suggesting that's all that needs to be reflected in the V3 API. If consider some of the problems we've had in Havana (Neutron port quota running out during server create, Errors from the neutron client being reported back as 500 errors in Nova) etc it seems to me that a lot of those would have been avoided if we had a much simpler model and pushed the cross-service complexity / convenience up to the client layer. For the associate and disassociate, the only convenience is being able to use the instance display name and security group name, which is already handled at the client layer. It seems a clearer case of duplicating what neutron offers. Agree that this is a clearer case - but in the context of V3 API changes I think we should look at the wider issue. Cheers, Phil ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] GPU passthrough support blueprints for OpenStack
Hi Brian, On the Hyper-V side we have Remote-Fx support in the works for Havana: https://blueprints.launchpad.net/nova/+spec/hyper-v-remotefx The code will be ready for review around next week. Thanks, Alessandro On Aug 14, 2013, at 01:07 , Brian Schott brian.sch...@nimbisservices.commailto:brian.sch...@nimbisservices.com wrote: Are there more recent blueprints related to adding GPU pass-through support? All that I can find are some stale blueprints that I created around the Cactus timeframe (while wearing a different hat) that are pretty out of date. I just heard a rumor that folks are doing Nvidia GRID K2 GPU passthrough with KVM successfully using linux 3.10.6 kernel with RHEL. In addition, Lorin and I did some GPU passthrough testing back in the spring with GRID K2 on HyperV, libvirt+xen, and XenServer. Slides are here: http://www.slideshare.net/bfschott/nimbis-schott-openstackgpustatus20130618 The virtualization support for GPU-enabled virtual desktops and GPGPU seems to have stabilized this year for server deployments. How is this going to be supported in OpenStack? Brian - Brian Schott, CTO Nimbis Services, Inc. brian.sch...@nimbisservices.com ph: 443-274-6064 fx: 443-274-6060 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] dnsmasq doesn't send DHCPACK
Dear list, I'm facing a problem related with network bring up inside a virtual instance Scientific Linux 5. My Openstack installation has been made with packstack (with options --allinone --os-swift-install=n --os-quantum-install=n --os-cinder-install=n), using the latest Grizzly, on a Scientific Linux 6.4 host machine: [root@lxadorigo log]# rpm -qa|egrep 'openstack|qemu|virt|dnsmasq' qemu-img-0.12.1.2-2.355.el6_4.6.x86_64 openstack-packstack-2013.1.1-0.24.dev660.el6.noarch dnsmasq-utils-2.48-13.el6.x86_64 libvirt-client-0.10.2-18.el6_4.9.x86_64 libvirt-python-0.10.2-18.el6_4.9.x86_64 openstack-nova-scheduler-2013.1.3-1.el6.noarch openstack-glance-2013.1.2-2.el6.noarch openstack-nova-common-2013.1.3-1.el6.noarch openstack-nova-network-2013.1.3-1.el6.noarch gpxe-roms-qemu-0.9.7-6.9.el6.noarch libvirt-0.10.2-18.el6_4.9.x86_64 openstack-nova-compute-2013.1.3-1.el6.noarch openstack-utils-2013.1-8.el6.noarch python-django-openstack-auth-1.0.7-1.el6.noarch openstack-nova-api-2013.1.3-1.el6.noarch qemu-kvm-0.12.1.2-2.355.el6_4.6.x86_64 openstack-nova-conductor-2013.1.3-1.el6.noarch openstack-nova-cert-2013.1.3-1.el6.noarch kernel-2.6.32-358.114.1.openstack.el6.gre.2.x86_64 virt-what-1.11-1.2.el6.x86_64 openstack-nova-novncproxy-2013.1.3-1.el6.noarch openstack-dashboard-2013.1.1-1.el6.noarch kernel-firmware-2.6.32-358.114.1.openstack.el6.gre.2.noarch openstack-keystone-2013.1.3-1.el6.noarch openstack-nova-console-2013.1.3-1.el6.noarch dnsmasq-2.48-13.el6.x86_64 [root@lxadorigo log]# uname -r 2.6.32-358.114.1.openstack.el6.gre.2.x86_64 When I upload a Scientific Linux 6 AMI image (linked to a proper kernel and initramfs) and then I start it, I do not have any problem with it; the instance starts and this is the response the dhcp client gets from the dnsmasq: Aug 14 13:48:43 lxadorigo dnsmasq[3321]: read /etc/hosts - 2 addresses Aug 14 13:48:43 lxadorigo dnsmasq[3321]: read /var/lib/nova/networks/nova-br100.conf Aug 14 13:48:45 lxadorigo kernel: kvm: 24134: cpu0 disabled perfctr wrmsr: 0xc1 data 0xabcd Aug 14 13:48:51 lxadorigo kernel: device vnet0 entered promiscuous mode Aug 14 13:48:51 lxadorigo kernel: br100: port 1(vnet0) entering forwarding state Aug 14 13:48:51 lxadorigo qemu-kvm: Could not find keytab file: /etc/qemu/krb5.tab: No such file or directory Aug 14 13:48:54 lxadorigo ntpd[2283]: Listening on interface #15 vnet0, fe80::fc16:3eff:fe16:fbbc#123 Enabled Aug 14 13:48:56 lxadorigo kernel: kvm: 24392: cpu0 unhandled wrmsr: 0x391 data 200f Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPDISCOVER(br100) fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPOFFER(br100) 192.168.32.4 fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPREQUEST(br100) 192.168.32.4 fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPACK(br100) 192.168.32.4 fa:16:3e:16:fb:bc sl64-5gb If I upload a Scientific Linux 5, when its dhclient asks the dnsmasq for negotiation this is the dnsmasq's answer: Aug 14 14:08:53 lxadorigo dnsmasq[25019]: read /etc/hosts - 2 addresses Aug 14 14:08:53 lxadorigo dnsmasq[25019]: read /var/lib/nova/networks/nova-br100.conf Aug 14 14:08:55 lxadorigo kernel: kvm: 26374: cpu0 disabled perfctr wrmsr: 0xc1 data 0xabcd Aug 14 14:09:00 lxadorigo kernel: device vnet0 entered promiscuous mode Aug 14 14:09:00 lxadorigo kernel: br100: port 1(vnet0) entering forwarding state Aug 14 14:09:00 lxadorigo qemu-kvm: Could not find keytab file: /etc/qemu/krb5.tab: No such file or directory Aug 14 14:09:03 lxadorigo ntpd[2283]: Listening on interface #18 vnet0, fe80::fc16:3eff:fea4:1cb8#123 Enabled Aug 14 14:09:12 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:12 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:20 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:20 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:28 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:28 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:35 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:35 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:42 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:42 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:56 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:56 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:10:07 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:10:07 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 without any DHCPOFFERS message. Now, I'm not a network and/or dnsmasq
Re: [openstack-dev] dnsmasq doesn't send DHCPACK
Hello, Alvise It is possible that the version of dnsmasq and lease time is an issue: https://bugs.launchpad.net/nova/+bug/887162 http://markmail.org/message/7kjf4hljszpydsrx#query:+page:1+mid:7kjf4hljszpydsrx+state:results Hope this helps. -- Best regards, Oleg Gelbukh Mirantis Inc. On Wed, Aug 14, 2013 at 4:19 PM, Alvise Dorigo alvise.dor...@pd.infn.itwrote: Dear list, I'm facing a problem related with network bring up inside a virtual instance Scientific Linux 5. My Openstack installation has been made with packstack (with options --allinone --os-swift-install=n --os-quantum-install=n --os-cinder-install=n), using the latest Grizzly, on a Scientific Linux 6.4 host machine: [root@lxadorigo log]# rpm -qa|egrep 'openstack|qemu|virt|dnsmasq' qemu-img-0.12.1.2-2.355.el6_4.6.x86_64 openstack-packstack-2013.1.1-0.24.dev660.el6.noarch dnsmasq-utils-2.48-13.el6.x86_64 libvirt-client-0.10.2-18.el6_4.9.x86_64 libvirt-python-0.10.2-18.el6_4.9.x86_64 openstack-nova-scheduler-2013.1.3-1.el6.noarch openstack-glance-2013.1.2-2.el6.noarch openstack-nova-common-2013.1.3-1.el6.noarch openstack-nova-network-2013.1.3-1.el6.noarch gpxe-roms-qemu-0.9.7-6.9.el6.noarch libvirt-0.10.2-18.el6_4.9.x86_64 openstack-nova-compute-2013.1.3-1.el6.noarch openstack-utils-2013.1-8.el6.noarch python-django-openstack-auth-1.0.7-1.el6.noarch openstack-nova-api-2013.1.3-1.el6.noarch qemu-kvm-0.12.1.2-2.355.el6_4.6.x86_64 openstack-nova-conductor-2013.1.3-1.el6.noarch openstack-nova-cert-2013.1.3-1.el6.noarch kernel-2.6.32-358.114.1.openstack.el6.gre.2.x86_64 virt-what-1.11-1.2.el6.x86_64 openstack-nova-novncproxy-2013.1.3-1.el6.noarch openstack-dashboard-2013.1.1-1.el6.noarch kernel-firmware-2.6.32-358.114.1.openstack.el6.gre.2.noarch openstack-keystone-2013.1.3-1.el6.noarch openstack-nova-console-2013.1.3-1.el6.noarch dnsmasq-2.48-13.el6.x86_64 [root@lxadorigo log]# uname -r 2.6.32-358.114.1.openstack.el6.gre.2.x86_64 When I upload a Scientific Linux 6 AMI image (linked to a proper kernel and initramfs) and then I start it, I do not have any problem with it; the instance starts and this is the response the dhcp client gets from the dnsmasq: Aug 14 13:48:43 lxadorigo dnsmasq[3321]: read /etc/hosts - 2 addresses Aug 14 13:48:43 lxadorigo dnsmasq[3321]: read /var/lib/nova/networks/nova-br100.conf Aug 14 13:48:45 lxadorigo kernel: kvm: 24134: cpu0 disabled perfctr wrmsr: 0xc1 data 0xabcd Aug 14 13:48:51 lxadorigo kernel: device vnet0 entered promiscuous mode Aug 14 13:48:51 lxadorigo kernel: br100: port 1(vnet0) entering forwarding state Aug 14 13:48:51 lxadorigo qemu-kvm: Could not find keytab file: /etc/qemu/krb5.tab: No such file or directory Aug 14 13:48:54 lxadorigo ntpd[2283]: Listening on interface #15 vnet0, fe80::fc16:3eff:fe16:fbbc#123 Enabled Aug 14 13:48:56 lxadorigo kernel: kvm: 24392: cpu0 unhandled wrmsr: 0x391 data 200f Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPDISCOVER(br100) fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPOFFER(br100) 192.168.32.4 fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPREQUEST(br100) 192.168.32.4 fa:16:3e:16:fb:bc Aug 14 13:49:04 lxadorigo dnsmasq-dhcp[3321]: DHCPACK(br100) 192.168.32.4 fa:16:3e:16:fb:bc sl64-5gb If I upload a Scientific Linux 5, when its dhclient asks the dnsmasq for negotiation this is the dnsmasq's answer: Aug 14 14:08:53 lxadorigo dnsmasq[25019]: read /etc/hosts - 2 addresses Aug 14 14:08:53 lxadorigo dnsmasq[25019]: read /var/lib/nova/networks/nova-br100.conf Aug 14 14:08:55 lxadorigo kernel: kvm: 26374: cpu0 disabled perfctr wrmsr: 0xc1 data 0xabcd Aug 14 14:09:00 lxadorigo kernel: device vnet0 entered promiscuous mode Aug 14 14:09:00 lxadorigo kernel: br100: port 1(vnet0) entering forwarding state Aug 14 14:09:00 lxadorigo qemu-kvm: Could not find keytab file: /etc/qemu/krb5.tab: No such file or directory Aug 14 14:09:03 lxadorigo ntpd[2283]: Listening on interface #18 vnet0, fe80::fc16:3eff:fea4:1cb8#123 Enabled Aug 14 14:09:12 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:12 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:20 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:20 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:28 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:28 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:35 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:35 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:42 lxadorigo dnsmasq-dhcp[25019]: DHCPDISCOVER(br100) fa:16:3e:a4:1c:b8 Aug 14 14:09:42 lxadorigo dnsmasq-dhcp[25019]: DHCPOFFER(br100) 192.168.32.2 fa:16:3e:a4:1c:b8 Aug 14 14:09:56
Re: [openstack-dev] [Ceilometer] Concerning get_resources/get_meters and the Ceilometer API
On 08/14/2013 06:17 AM, Julien Danjou wrote: On Tue, Aug 13 2013, Thomas Maddox wrote: Hi Thomas, * Driving get_resources() with the Meter table instead of the Resource table. This is mainly because of the additional filtering available in the Meter table, which allows us to satisfy a use case like getting a list of resources a user had during a period of time to get meters to compute billing with. The semantics are tripping me up a bit; the question this boiled down to for me was: why use a resource query to get meters to show usage by a tenant? I was curious about why we needed the timestamp filtering when looking at Resources, and why we would use Resource as a way to get at metering data, rather than a Meter request itself? This was answered by resources being the current vector to get at metering data for a tenant in terms of resources, if I understood correctly. * With this implementation, we have to do aggregation to get at the discrete Resources (via the Meter table) rather than just filtering the already distinct ones in the Resource table. I think I already answered that in a previous email when I said drop the resource table. :) * This brought up some confusion with the API for me with the major use cases I can think of: * As a new consumer of this API, I would think that /resource/resource_id would get me details for a resource, e.g. current state, when it was created, last updated/used timestamp, who owns it; not the attributes from the first sample to come through about it s/first/last/ actually I wouldn't disagree if you had such improvements to propose. However, we're pretty flexible in Ceilometer as we don't allow only metering OpenStack; be careful of somethings like who owns it might change in other systems than OpenStack for example, depending on the time range you filter on. * I would think that /meter/?q.field=resource_idq.value=resource_id ought to get me a list of meter(s) details for a specific resource, e.g. name, unit, and origin; but not a huge mixture of samples. That could be a nice improvement indeed. * Additionally /meter/?q.field=user_idq.value=user_id would get me a list of all meters that are currently related to the user Same as above. * The ultimate use case, for billing queries, I would think that /meter/meter_id/statistics?time filtersuser(resource_id) would get me the measurements for that meter to bill for. We'd like that too, though it's not always perfect since we don't handle the different counter types explicitly. If I understand correctly, one main intent driving this is wanting to avoid end users having to write a bunch of API requests themselves from the billing side and instead just drill down from payloads for each resource to get the billing information for their customers. It also looks like there's a BP to add grouping functionality to statistics queries to allow us this functionality easily (this one, I think: https://blueprints.launchpad.net/ceilometer/+spec/api-group-by). Yes. I'm new to this project, so I'm trying to get a handle on how we got here and maybe offer some outside perspective, if it's needed or wanted. =] I think you got the picture right. We're trying to improve the API, but we always happy to get help. There's a sort of meta blueprint: https://blueprints.launchpad.net/ceilometer/+spec/api-v2-improvement With various ideas to improve the API. It's assigned to me, though I didn't implement most of the ideas there, and won't probably have time to implement them all, so feel free to contribute! +1 ... I think that would clear up a lot of confusion not only in the api, but the underlying db models. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Neutron] Ceilometer support
Hi Gary, Isn't it something that 'metering support' is going to solve? A couple of patches addressing this are on review. Thanks, Eugene. On Wed, Aug 14, 2013 at 3:47 PM, Gary Kotton gkot...@vmware.com wrote: [image: Description: https://fbcdn-sphotos-f-a.akamaihd.net/hphotos-ak-prn2/1098074_725247340838055_1378530957_n.jpg] ** ** ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev image001.jpg___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
On 08/13/2013 06:20 PM, Dolph Mathews wrote: With regard to: https://blueprints.launchpad.net/keystone/+spec/key-distribution-server During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. I share the same concern, especially considering the API only recently went up for formal review [3], and the WIP implementation is still failing smokestack [4]. Since KDS is a security tightening in acase where there is no security at all, adding it in can only improve security. It is a relatively simple extension from the keystone side. THe corresponding change is in the client, and that has already merged. I'm happy to see the reviews in question continue to receive their fair share of attention over the next few weeks, but can (and should?) merging be delayed until icehouse while more security-focused eyes have time to review the code? Ceilometer and nova would both be affected by a delay, as both have use cases for consuming trusted messaging [5] (a dependency of the bp in question). Thanks for you feedback! [1]: http://eavesdrop.openstack.org/irclogs/%23openstack-meeting/%23openstack-meeting.2013-08-13.log [2]: http://paste.openstack.org/raw/44075/ [3]: https://review.openstack.org/#/c/40692/ [4]: https://review.openstack.org/#/c/37118/ [5]: https://blueprints.launchpad.net/oslo/+spec/trusted-messaging ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] GPU passthrough support blueprints for OpenStack
Thanks for the pointers. I'll try to work with the UCS-ISI team that currently maintains those stories and see if we can reconcile or close the old blueprints. - Brian Schott, CTO Nimbis Services, Inc. brian.sch...@nimbisservices.com ph: 443-274-6064 fx: 443-274-6060 On Aug 14, 2013, at 6:03 AM, Bob Ball bob.b...@citrix.com wrote: Hi Brian, Instead of specific GPU pass-through there are blueprints for generic PCI pass through at https://blueprints.launchpad.net/nova/+spec/pci-passthrough-base - these are being worked on for Havana and we're hopeful they will get in (possibly libvirt only) In terms of vGPU rather than whole-GPU pass through, the slides are correct that XenServer support is very nearly here: http://www.xenserver.org/discuss-virtualization/q-and-a/nvidia-vgpu-on-xenserver-eg-gpu-hypervisor.html Support for static allocation of vGPU is likely going to be the first phase - and adding it as a resource similar to RAM/CPU isn't likely to be proposed until a second phase. As such, I think the integration should be very similar to PCI pass through above. Bob -Original Message- From: Brian Schott [mailto:brian.sch...@nimbisservices.com] Sent: 13 August 2013 23:07 To: openstack-dev@lists.openstack.org Subject: [openstack-dev] GPU passthrough support blueprints for OpenStack Are there more recent blueprints related to adding GPU pass-through support? All that I can find are some stale blueprints that I created around the Cactus timeframe (while wearing a different hat) that are pretty out of date. I just heard a rumor that folks are doing Nvidia GRID K2 GPU passthrough with KVM successfully using linux 3.10.6 kernel with RHEL. In addition, Lorin and I did some GPU passthrough testing back in the spring with GRID K2 on HyperV, libvirt+xen, and XenServer. Slides are here: http://www.slideshare.net/bfschott/nimbis-schott- openstackgpustatus20130618 The virtualization support for GPU-enabled virtual desktops and GPGPU seems to have stabilized this year for server deployments. How is this going to be supported in OpenStack? Brian - Brian Schott, CTO Nimbis Services, Inc. brian.sch...@nimbisservices.com ph: 443-274-6064 fx: 443-274-6060 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev signature.asc Description: Message signed with OpenPGP using GPGMail ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
On Tue, Aug 13, 2013 at 5:54 PM, Simo Sorce s...@redhat.com wrote: On Tue, 2013-08-13 at 17:20 -0500, Dolph Mathews wrote: With regard to: https://blueprints.launchpad.net/keystone/+spec/key-distribution-server Well I am of course biased so take my comments with a grain of salt, that said... During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. Is ttx going to review any security implications ? The code does not mature just because is sit there untouched for more or less time. I share the same concern, especially considering the API only recently went up for formal review [3], While the API may be important it has little to no bearing over the security properties of the underlying code and mechanism. The document to review to understand and/or criticize the security implications is this: https://wiki.openstack.org/wiki/MessageSecurity and it has been available for quite a few months. and the WIP implementation is still failing smokestack [4]. This is a red herring, unfortunately Smokestack doesn't say why it is failing but we suppose it is due to something python 2.6 doesn't like (only the centos machine fails). I have been developing on 2.7 and was planning to do a final test on a machine with 2.6 once I had reviews agreeing no more fundamental changes were needed. My mistake - glancing through the patchset history I thought it was SmokeStack that was regularly failing, but it appears to be mostly Jenkins failures with SmokeStack failing most recently. Either way, I try to avoid reviewing code that is still failing automated tests, so I have yet to review the KDS implementation at all. I'm happy to see the reviews in question continue to receive their fair share of attention over the next few weeks, but can (and should?) merging be delayed until icehouse while more security-focused eyes have time to review the code? I would agree to this only if you can name individuals that are going to do a security review, otherwise I see no real reason to delay, as it will cost time to keep patches up to date, and I'd rather not do that if no one is lining up to do a security review. keystone-core, at least... that's part of our responsibility. The commit message also lacks a SecurityImpact tag. FWIW I did circulate the design for the security mechanism internally in Red Hat to some people with some expertise in crypto matters. I'd love to see their feedback provided publicly. Simo. -- Simo Sorce * Red Hat, Inc * New York -- -Dolph ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] Question about get_meters query using a JOIN
On 8/14/13 3:26 AM, Julien Danjou jul...@danjou.info wrote: On Tue, Aug 13 2013, Thomas Maddox wrote: I was curious about why we went for a JOIN here rather than just using the meter table initially? https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/im pl_sqlalchemy.py#L336-L391. Doug had mentioned that some performance testing had gone on with some of these queries, so before writing up requests to change this to the meter table only, I wanted to see if this was a result of that performance testing? Like the JOIN was less expensive than a DISTINCT. Because the SQL driver has been historically a straight conversion from MongoDB, who was working that way. The MongoDB driver had these 2 collections, meter and resources, and used both of them to construct these data. The original idea was to register resources in the eponym table, and use it to filter. Unfortunately, that didn't work out, especially with APIv2 that allows to filter on many things. Among other things, there was timestamp fields in the resource table that were used to filter based on timestamp, but this was a naive implementation that failed pretty soon, and we removed and replace with a more solid one based on the meter table. What you see here are the rest of that time, and in the end, this resource table should be removed AFAICT, in MongoDB also. Ahhh yep, when I look at the two side-by-side I can see what you're saying. Thanks for the explanation! So, then should I write up a BP for improvements of this sort, like the API improvement one? It seems like we may get better results if the SQL implementation is more SQL-like than Mongo-like since they are fundamentally different things. I noticed some notes in there also about the JOIN being especially expensive. -- Julien Danjou ;; Free Software hacker ; freelance consultant ;; http://julien.danjou.info ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
Simo Sorce wrote: During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. Is ttx going to review any security implications ? The code does not mature just because is sit there untouched for more or less time. This is me wearing my vulnerability management hat on. The trick is that we (the VMT) have to support security issues for code that will be shipped in stable/havana. The most embarrassing security issues we had in the past were with code that didn't see a fair amount of time in master before we had to start supporting it. So for us there is a big difference between landing the KDS now and have it security-supported after one month of usage, and landing it in a few weeks and have it security-supported after 7 months of usage. After 7 months I'm pretty sure most of the embarrassing issues will be ironed out. I don't really want us to repeat the mistakes of the past where we shipped really new code in keystone that ended up not really usable, but which we still had to support security-wise due to our policy. By security implications, I mean that this is a domain (like, say, token expiration) where even basic bugs can easily create a vulnerability. We just don't have the bandwidth to ship an embargoed security advisory for every bug that will be found in the KDS one month from now. I would agree to this only if you can name individuals that are going to do a security review, otherwise I see no real reason to delay, as it will cost time to keep patches up to date, and I'd rather not do that if no one is lining up to do a security review. FWIW I did circulate the design for the security mechanism internally in Red Hat to some people with some expertise in crypto matters. Are you saying it won't have significantly less issues in 7 months just by the virtue of being landed in master and put into use in various projects ? Or that it was so thoroughly audited that my fears are unwarranted ? -- Thierry Carrez (ttx) ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] Concerning get_resources/get_meters and the Ceilometer API
On 8/14/13 10:29 AM, Julien Danjou jul...@danjou.info wrote: On Wed, Aug 14 2013, Thomas Maddox wrote: Am I misunderstanding the code? This looks like it's returning the first sample's details: https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/im pl _mongodb.py#L578-L584. When I change the metadata attribute's aggregation function from $first to $last, I get the latest state of the resource, which corrects the bug I'm working on. Otherwise, a newly built instance sits in a 'scheduling' state, according to the API call (https://bugs.launchpad.net/ceilometer/+bug/1208547). Haha! I'm pretty sure it used to be last. The aggregate() function is recent, so that may be a regression that we didn't catch. Anyway the intention is last, not first. I blame missing tests! Lol, understood. I'll roll that in with my bug fix since it's related. =] That's definitely a good point; I didn't know that. I suppose if we wanted to make this API change, it'd have to be 'who owns it currently' as part of the contract for what details are returned. The event body or samples can give the historical details when desired. From a billing perspective, it'd be good to know ownership over time in order to bill appropriately for binary ownership billing rather than usage. H... YesŠ I admit we stuck to very simple case and assumptions in the API. There's a lot of corner case we aren't handling correctly, but it never mattered so far. We're trying to get better at it. As I already stated at some point, we need more test for this corner case and more fixes. :) Cheers! -Thomas -- Julien Danjou # Free Software hacker # freelance consultant # http://julien.danjou.info ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [Glance] Blueprint proposal - Import / Export images with user properties
Hi, I would like to discuss here about two blueprint proposal (maybe could I merge them into one if you prefer) : https://blueprints.launchpad.net/glance/+spec/api-v2-export-properties https://blueprints.launchpad.net/glance/+spec/api-v2-import-properties *Use case* : I would like to set specific properties to an image which could represent a signature, and useful for licensing requirements for example. To do that, I should be able to export an image with user properties included. Then, a user could reuse the exported image in the public cloud, and Glance will be aware about its properties. Obviously, we need the import / export feature. The idea here is to be able to identify an image after cloning or whatever with a property field. Of course, the user could break it in editing the image manually, but I consider he / she won't. Let me know if you have any thoughts and if the blueprint is valuable. Regards, -- Emilien Macchi # OpenStack Engineer // eNovance Inc. http://enovance.com // ? emil...@enovance.com ? +33 (0)1 49 70 99 80 // 10 rue de la Victoire 75009 Paris signature.asc Description: OpenPGP digital signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
Dolph Mathews wrote: With regard to: https://blueprints.launchpad.net/keystone/+spec/key-distribution-server [...] Dolph: you don't mention Barbican at all, does that mean that the issue is settled and the KDS should live in keystone ? A side-benefit of landing early in Icehouse rather than late in Havana, IMHO, was that we could put everyone in the same room (and around the same beers) and get a single path forward. I'm a bit worried (that's with my release management hat on) that if everyone discovers that Barbican is the way to go for key distribution at the Hong-Kong summit, we would now have to deprecate the in-Keystone KDS over several releases just because it landed a few weeks too early. That said I haven't followed closely the latest discussions on this, so maybe this is not as much duplication of effort as I think ? -- Thierry Carrez (ttx) ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] Question about get_meters query using a JOIN
On Wed, Aug 14 2013, Thomas Maddox wrote: Ahhh yep, when I look at the two side-by-side I can see what you're saying. Thanks for the explanation! So, then should I write up a BP for improvements of this sort, like the API improvement one? It seems like we may get better results if the SQL implementation is more SQL-like than Mongo-like since they are fundamentally different things. I noticed some notes in there also about the JOIN being especially expensive. Honestly, I wouldn't require a blueprint for that but if you prefer writing one before because you think it's going to take a while, be split in a lot of patches, go ahead. At this stage of development anyway we're not going to wait on it. :) -- Julien Danjou # Free Software hacker # freelance consultant # http://julien.danjou.info signature.asc Description: PGP signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] Create a tar.gz stream from a block device
Hi Stackers, I would like to create a readable tar.gz stream, inside that stream I want to see only one file, and that file's content should be read from a device. The virtual disk is attached to domU as a block device (/dev/xvdb) I would like to produce a tar.gz, that contains the contents of /dev/xvdb, file name inside the tar.gz does not matter. Here is my current solution: https://review.openstack.org/41651 The main problem is, that tarfile does not support adding content incrementaly, it only has a addfile method: http://docs.python.org/2/library/tarfile.html tfile = tarfile.open(fileobj=output, mode='w|gz') ... tfile.addfile(tinfo, fileobj=input_file) And for the glance client, I need to provide a readable file-like: image_service.update(self.context, image_id, metadata, image_stream, purge_props=False) The change is solving the issue by creating a separate process to create the targz stream, and give that process' output stream to glance. Any other ideas? Many thanks, -- Mate Lakat ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Migrating to testr parallel in tempest
On 2013-08-13 16:39, Clark Boylan wrote: On Tue, Aug 13, 2013 at 1:25 PM, Matthew Treinish mtrein...@kortar.org wrote: Hi everyone, So for the past month or so I've been working on getting tempest to work stably with testr in parallel. As part of this you may have noticed the testr-full jobs that get run on the zuul check queue. I was using that job to debug some of the more obvious race conditions and stability issues with running tempest in parallel. After a bunch of fixes to tempest and finding some real bugs in some of the projects things seem to have smoothed out. So I pushed the testr-full run to the gate queue earlier today. I'll be keeping track of the success rate of this job vs the serial job and use this as the determining factor before we push this live to be the default for all tempest runs. So assuming that the success rate matches up well enough with serial job on the gate queue then I will push out the change that will migrate all the voting jobs to run in parallel hopefully either Friday afternoon or early next week. Also, if anyone has any input on what threshold they feel is good enough for this I'd welcome any input on that. For example, do we want to ensure a = 1:1 match for job success? Or would something like 90% as stable as the serial job be good enough considering the speed advantage. (The parallel runs take about half as much time as a full serial run, the parallel job normally finishes in ~25-30min) Since this affects almost every project I don't want to define this threshold without input from everyone. After there is some more data for the gate queue's parallel job I'll have some pretty graphite graphs that I can share comparing the success trends between the parallel and serial jobs. So at this point we're in the home stretch and I'm asking for everyone's help in getting this merged. So, if everyone who is reviewing and pushing commits could watch the results from these non-voting jobs and if things fail on the parallel job but not the serial job please investigate the failure and open a bug if necessary. If it turns out to be a bug in tempest please link it against this blueprint: https://blueprints.launchpad.net/tempest/+spec/speed-up-tempest so that I'll give it the attention it deserves. I'd hate to get this close to getting this merged and have a bit of racy code get merged at the last second and block us for another week or two. I feel that we need to get this in before the H3 rush starts up as it will help everyone get through the extra review load faster. Getting this in before the H3 rush would be very helpful. When we made the switch with Nova's unittests we fixed as many of the test bugs that we could find, merged the change to switch the test runner, then treated all failures as very high priority bugs that received immediate attention. Getting this in before H3 will give everyone a little more time to debug any potential new issues exposed by Jenkins or people running the tests locally. I think we should be bold here and merge this as soon as we have good numbers that indicate the trend is for these tests to pass. Graphite can give us the pass to fail ratios over time, as long as these trends are similar for both the old nosetest jobs and the new testr job I say we go for it. (Disclaimer: most of the projecst I work on are not affected by the tempest jobs; however, I am often called upon to help sort out issues in the gate). I'm inclined to agree. It's not as if we don't have transient failures now, and if we're looking at a 50% speedup in recheck/verify times then as long as the new version isn't significantly less stable it should be a net improvement. Of course, without hard numbers we're kind of discussing in a vacuum here. -Ben ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
It's just an extension, shouldn't be treated differently as long as it follow the rules and regulations. 1. Bp 2. Spec (identity-api) 3. Server-side changes (keystone) 4. Client-side changes if any (python-keystoneclient) If OpenStack security community is participating in the code reviews, that would even be awesomer. Guang From: Adam Young [mailto:ayo...@redhat.com] Sent: Wednesday, August 14, 2013 6:24 AM To: openstack-dev@lists.openstack.org Subject: Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse? On 08/13/2013 06:20 PM, Dolph Mathews wrote: With regard to: https://blueprints.launchpad.net/keystone/+spec/key-distribution-server During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. I share the same concern, especially considering the API only recently went up for formal review [3], and the WIP implementation is still failing smokestack [4]. Since KDS is a security tightening in acase where there is no security at all, adding it in can only improve security. It is a relatively simple extension from the keystone side. THe corresponding change is in the client, and that has already merged. I'm happy to see the reviews in question continue to receive their fair share of attention over the next few weeks, but can (and should?) merging be delayed until icehouse while more security-focused eyes have time to review the code? Ceilometer and nova would both be affected by a delay, as both have use cases for consuming trusted messaging [5] (a dependency of the bp in question). Thanks for you feedback! [1]: http://eavesdrop.openstack.org/irclogs/%23openstack-meeting/%23openstack-mee ting.2013-08-13.log [2]: http://paste.openstack.org/raw/44075/ [3]: https://review.openstack.org/#/c/40692/ [4]: https://review.openstack.org/#/c/37118/ [5]: https://blueprints.launchpad.net/oslo/+spec/trusted-messaging ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev smime.p7s Description: S/MIME cryptographic signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Swift, netifaces, PyPy, and cffi
On 2013-08-13 16:58, Alex Gaynor wrote: One of the issues that came up in this review however, is that cffi is not packaged in the most recent Ubuntu LTS (and likely other distributions), although it is available in raring, and in a PPA (http://packages.ubuntu.com/raring/python-cffi [2] nd https://launchpad.net/~pypy/+archive/ppa?field.series_filter=precise [3] respectively). As a result of this, we wanted to get some feedback on which direction is best to go: a) cffi-only approach, this is obviously the simplest approach, and works everywhere (assuming you can install a PPA, use pip, or similar for cffi) b) wait until the next LTS to move to this approach (requires waiting until 2014 for PyPy support) c) Support using either netifaces or cffi: most complex, and most code, plus one or the other dependencies aren't well supported by most tools as far as I know. It doesn't appear to me that this is available for RHEL yet, although it looks like they're working on it: https://admin.fedoraproject.org/updates/python-cffi-0.6-4.el6 That's also going to need to happen before we can do this, I think. -Ben ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] Pagination
So, Are we saying that UIs built on OpenStack APIs shouldn't be able to show traditional pagination controls? Or am I missing how this should work with marker/limit? e.g. for 11 pages of content, something like: 1 2 3 .. 10 11 Thanks, Kiall On 13/08/13 22:45, Jay Pipes wrote: On 08/13/2013 05:04 PM, Gabriel Hurley wrote: I have been one of the earliest, loudest, and most consistent PITA's about pagination, so I probably oughta speak up. I would like to state three facts: 1. Marker + limit (e.g. forward-only) pagination is horrific for building a user interface. 2. Pagination doesn't scale. 3. OpenStack's APIs have historically had useless filtering capabilities. In a world where pagination is a must-have feature we need to have page number + limit pagination in order to build a reasonable UI. Ironically though, I'm in favor of ditching pagination altogether. It's the lowest-common denominator, used because we as a community haven't buckled down and built meaningful ways for our users to get to the data they really want. Filtering is great, but it's only 1/3 of the solution. Let me break it down with problems and high level solutions: Problem 1: I know what I want and I need to find it. Solution: filtering/search systems. This is a good place to start. Glance has excellent filtering/search capabilities -- built in to the API from early on in the Essex timeframe, and only expanded over the last few releases. Pagination solutions should build on a solid filtering/search functionality in the API, where there is a consistent sort key and direction (either hard-coded or user-determined, doesn't matter). Limit/offset pagination solutions (forward and backwards paging, random skip-to-a-page) are inefficient from a SQL query perspective and should be a last resort, IMO, compared to limit/marker. With some smart session-storage of a page's markers, backwards paging with limit/marker APIs is certainly possible -- just store the previous page's marker. Problem 2: I don't know what I want, and it may or may not exist. Solution: tailored discovery mechanisms. This should not be a use case that we spend much time on. Frankly, this use case can be summarized as the window shopper scenario. Providing a quality search/filtering mechanism, including the *API* itself providing REST-ful discovery of the filters and search criteria the API supports, is way more important... Problem 3: I need to know something about *all* the data in my system. Solution: reporting systems. Sure, no disagreement there. We've got the better part of none of that. I disagree. Some of the APIs have support for a good bit of search/filtering. We just need to bring all the projects up to search speed, Captain. Best, -jay p.s. I very often go to the second and third pages of Google searches. :) But I never skip to the 127th page of results. But I'd like to solve these issues. I have lots of thoughts on all of those, and I think the UX and design communities can offer a lot in terms of the usability of the solutions we come up with. Even more, I think this would be an awesome working group session at the next summit to talk about nothing other than how can we get rid of pagination? As a parting thought, what percentage of the time do you click to the second page of results in Google? All the best, - Gabriel ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Glance] Blueprint proposal - Import / Export images with user properties
Lets dig into this a bit more so that I can understand it. Given that we have properties that we want to export with an image, where would those properties be stored? Somewhere in the image data itself? I believe some image formats support metadata, but I can't imagine all of them would. Is there a specific format you're thinking of using? On Wed, Aug 14, 2013 at 8:36 AM, Emilien Macchi emilien.mac...@enovance.com wrote: Hi, I would like to discuss here about two blueprint proposal (maybe could I merge them into one if you prefer) : https://blueprints.launchpad.net/glance/+spec/api-v2-export-properties https://blueprints.launchpad.net/glance/+spec/api-v2-import-properties *Use case* : I would like to set specific properties to an image which could represent a signature, and useful for licensing requirements for example. To do that, I should be able to export an image with user properties included. Then, a user could reuse the exported image in the public cloud, and Glance will be aware about its properties. Obviously, we need the import / export feature. The idea here is to be able to identify an image after cloning or whatever with a property field. Of course, the user could break it in editing the image manually, but I consider he / she won't. Let me know if you have any thoughts and if the blueprint is valuable. Regards, -- Emilien Macchi # OpenStack Engineer // eNovance Inc. http://enovance.com // ✉ emil...@enovance.com ☎ +33 (0)1 49 70 99 80 // 10 rue de la Victoire 75009 Paris ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Skipping tests in tempest via config file
I am running tempest on OEL 6.3 (aka RHEL 6.3) and I had issues with python 2.6 and skipException[3], so now I'm using python 2.7 just for tempest. I also had difficulty with yum and python module dependency and made the transition to venv. This has reduced the yum dependency nightmare greatly. now that testr is default for tempest. testr does not appear to support --exclusion[1] or --stop[2]. I have a work around for --exclusion, by: testr list-tests | egrep -v regex-exclude-list unit-tests.txt testr --load-list unit-tests.txt I do not have a work around for --stop. [1]https://bugs.launchpad.net/testrepository/+bug/1208610 [2]https://bugs.launchpad.net/testrepository/+bug/1211926 [3]https://bugs.launchpad.net/tempest/+bug/1202815 On Tue, Aug 13, 2013 at 7:25 PM, Matt Riedemann mrie...@us.ibm.com wrote: I have the same issue. I run a subset of the tempest tests via nose on a RHEL 6.4 VM directly against the site-packages (not using virtualenv). I'm running on x86_64, ppc64 and s390x and have different issues on all of them (a mix of DB2 on x86_64 and MySQL on the others, and different nova/cinder drivers on each). What I had to do was just make a nose.cfg for each of them and throw that into ~/ for each run of the suite. The switch from nose to testr hasn't impacted me because I'm not using a venv. However, there was a change this week that broke me on python 2.6 and I opened this bug: *https://bugs.launchpad.net/tempest/+bug/1212071*https://bugs.launchpad.net/tempest/+bug/1212071 Thanks, *MATT RIEDEMANN* Advisory Software Engineer Cloud Solutions and OpenStack Development -- *Phone:* 1-507-253-7622 | *Mobile:* 1-507-990-1889* E-mail:* *mrie...@us.ibm.com* mrie...@us.ibm.com [image: IBM] 3605 Hwy 52 N Rochester, MN 55901-1407 United States From:Ian Wienand iwien...@redhat.com To:openstack-dev@lists.openstack.org, Date:08/13/2013 09:13 PM Subject:[openstack-dev] Skipping tests in tempest via config file -- Hi, I proposed a change to tempest that skips tests based on a config file directive [1]. Reviews were inconclusive and it was requested the idea be discussed more widely. Of course issues should go upstream first. However, sometimes test failures are triaged to a local/platform problem and it is preferable to keep everything else running by skipping the problematic tests while its being worked on. My perspective is one of running tempest in a mixed CI environment with RHEL, Fedora, etc. Python 2.6 on RHEL doesn't support testr (it doesn't do the setUpClass calls required by temptest) and nose upstream has some quirks that make it hard to work with the tempest test layout [2]. Having a common place in the temptest config to set these skips is more convienent than having to deal with the multiple testing environments. Another proposal is to have a separate JSON file of skipped tests. I don't feel strongly but it does seem like another config file. -i [1] https://review.openstack.org/#/c/39417/ [2] https://github.com/nose-devs/nose/pull/717 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev image/gif___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] SQL meter table schema improvements
On Wed, Aug 14 2013, Jay Pipes wrote: I submitted a bug report to improve the schema of the meter table to reduce duplication: https://bugs.launchpad.net/ceilometer/+bug/1211985 This would only affect the SQL storage driver. Interested in hearing thoughts on the above recommendations, which are: 1) Replace counter_type with an integer column counter_type_id that references a lookup table (counter_type) 2) Replace counter_unit with an integer column counter_unit_id that references a lookup table (counter_unit) 3) Replace counter_name with an integer column counter_id that references a lookup table (counter) Just those three changes would reduce the overall row size for records in the meter table by about 10%, by my calculations, which is a good improvement for a relatively painless set of modifications. That means we'll have to do 3 joins most of the time. Is that less costly than fetching these right away from the record? -- Julien Danjou -- Free Software hacker - freelance consultant -- http://julien.danjou.info signature.asc Description: PGP signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] Pagination
On 08/14/2013 12:25 PM, Mac Innes, Kiall wrote: So, Are we saying that UIs built on OpenStack APIs shouldn't be able to show traditional pagination controls? Or am I missing how this should work with marker/limit? No, not quite what I'm saying. The operation to get the total number of pages -- or more explicitly, the operation to get the *exact* number of pages in a list result -- is expensive, and in order to be reasonably efficient, some level of caching is almost always needed. However, being able to page forwards and backwards is absolutely possible with limit/marker solutions. It simply requires the paging client (in this case, Horizon), to store the list of previously seen page links returned in listing results (there is a next and prev link in the returned list images results, for example). e.g. for 11 pages of content, something like: 1 2 3 .. 10 11 Yeah, that's not an efficient operation unless you have some sort of caching in place. You can use things like MySQL's SQL_CALC_FOUND_ROWS, but that is not efficient since instead of stopping the query after LIMIT rows, you end up executing the entire query to determine the number of rows that *would* have been returned if no LIMIT was applied. In order to make such a thing efficient, you'd want to cache the value of SQL_CALC_FOUND_ROWS in the session and use that to calculate the number of pages. It's something that can be done, but isn't, IMHO, worth it to get the traditional UI you describe. Instead, a good filter/search UI would be better, with just next/prev links. Best, -jay Thanks, Kiall On 13/08/13 22:45, Jay Pipes wrote: On 08/13/2013 05:04 PM, Gabriel Hurley wrote: I have been one of the earliest, loudest, and most consistent PITA's about pagination, so I probably oughta speak up. I would like to state three facts: 1. Marker + limit (e.g. forward-only) pagination is horrific for building a user interface. 2. Pagination doesn't scale. 3. OpenStack's APIs have historically had useless filtering capabilities. In a world where pagination is a must-have feature we need to have page number + limit pagination in order to build a reasonable UI. Ironically though, I'm in favor of ditching pagination altogether. It's the lowest-common denominator, used because we as a community haven't buckled down and built meaningful ways for our users to get to the data they really want. Filtering is great, but it's only 1/3 of the solution. Let me break it down with problems and high level solutions: Problem 1: I know what I want and I need to find it. Solution: filtering/search systems. This is a good place to start. Glance has excellent filtering/search capabilities -- built in to the API from early on in the Essex timeframe, and only expanded over the last few releases. Pagination solutions should build on a solid filtering/search functionality in the API, where there is a consistent sort key and direction (either hard-coded or user-determined, doesn't matter). Limit/offset pagination solutions (forward and backwards paging, random skip-to-a-page) are inefficient from a SQL query perspective and should be a last resort, IMO, compared to limit/marker. With some smart session-storage of a page's markers, backwards paging with limit/marker APIs is certainly possible -- just store the previous page's marker. Problem 2: I don't know what I want, and it may or may not exist. Solution: tailored discovery mechanisms. This should not be a use case that we spend much time on. Frankly, this use case can be summarized as the window shopper scenario. Providing a quality search/filtering mechanism, including the *API* itself providing REST-ful discovery of the filters and search criteria the API supports, is way more important... Problem 3: I need to know something about *all* the data in my system. Solution: reporting systems. Sure, no disagreement there. We've got the better part of none of that. I disagree. Some of the APIs have support for a good bit of search/filtering. We just need to bring all the projects up to search speed, Captain. Best, -jay p.s. I very often go to the second and third pages of Google searches. :) But I never skip to the 127th page of results. But I'd like to solve these issues. I have lots of thoughts on all of those, and I think the UX and design communities can offer a lot in terms of the usability of the solutions we come up with. Even more, I think this would be an awesome working group session at the next summit to talk about nothing other than how can we get rid of pagination? As a parting thought, what percentage of the time do you click to the second page of results in Google? All the best, - Gabriel ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
Adam Young wrote: On 08/13/2013 06:20 PM, Dolph Mathews wrote: During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. I share the same concern, especially considering the API only recently went up for formal review [3], and the WIP implementation is still failing smokestack [4]. Since KDS is a security tightening in acase where there is no security at all, adding it in can only improve security. It's not really a question of more security or less security... It's about putting young sensitive code into a release, with the risk of having to issue a lot of security advisories for early bugs. I'm all for that code to land in the icehouse master branch as soon as it opens and that it gets put into good use by projects throughout the icehouse development cycle. I just think the benefits of waiting outweigh the benefits of landing it now. I explained why I prefer it to land in a few weeks rather than now... Can someone explain why they prefer the reverse ? Why does it have to be in havana ? -- Thierry Carrez (ttx) ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Swift, netifaces, PyPy, and cffi
I just chatted with the Python product owner at Red Hat, he says this is going to make it's way to the next step later today (this past weekend was a Fedora conference), so this should be happening soon. Joe: Yup, I'm familiar with that piece (I had lunch with Vish the other week and he's the one who suggested Swift as the best place to get started with OpenStack + PyPy). For those who don't know I'm one of the core developers of PyPy :) Alex On Wed, Aug 14, 2013 at 9:24 AM, Ben Nemec openst...@nemebean.com wrote: On 2013-08-13 16:58, Alex Gaynor wrote: One of the issues that came up in this review however, is that cffi is not packaged in the most recent Ubuntu LTS (and likely other distributions), although it is available in raring, and in a PPA (http://packages.ubuntu.com/**raring/python-cffihttp://packages.ubuntu.com/raring/python-cffi[2] nd https://launchpad.net/~**pypy/+archive/ppa?field.** series_filter=precisehttps://launchpad.net/~pypy/+archive/ppa?field.series_filter=precise [3] respectively). As a result of this, we wanted to get some feedback on which direction is best to go: a) cffi-only approach, this is obviously the simplest approach, and works everywhere (assuming you can install a PPA, use pip, or similar for cffi) b) wait until the next LTS to move to this approach (requires waiting until 2014 for PyPy support) c) Support using either netifaces or cffi: most complex, and most code, plus one or the other dependencies aren't well supported by most tools as far as I know. It doesn't appear to me that this is available for RHEL yet, although it looks like they're working on it: https://admin.fedoraproject.** org/updates/python-cffi-0.6-4.**el6https://admin.fedoraproject.org/updates/python-cffi-0.6-4.el6 That's also going to need to happen before we can do this, I think. -Ben __**_ OpenStack-dev mailing list OpenStack-dev@lists.openstack.**org OpenStack-dev@lists.openstack.org http://lists.openstack.org/**cgi-bin/mailman/listinfo/**openstack-devhttp://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev -- I disapprove of what you say, but I will defend to the death your right to say it. -- Evelyn Beatrice Hall (summarizing Voltaire) The people's good is the highest law. -- Cicero GPG Key fingerprint: 125F 5C67 DFE9 4084 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [ANVIL] Missing openvswitch dependency for basic-neutron.yaml persona
Hi Sylvain, Thanks for checking looking it over. I think another anvil dev has started looking at the bug but here is an explanation of the spec process anyway :-) Preparing: which does the creation of the SRPMs 0. https://github.com/stackforge/anvil/blob/master/anvil/actions/prepare.py#L71 a. This then creates a yum class (which will do the work, this was abstracted so that its relatively easy to say have deb packaging later if ever wanted…) 1. https://github.com/stackforge/anvil/blob/master/anvil/packaging/yum.py#L150 then gets called will either package the openstack dependencies or the openstack component itself. a. Lets go down the openstack component itself path for now (the other path is simpler). 2. https://github.com/stackforge/anvil/blob/master/anvil/packaging/yum.py#L572 gets called a. This will create the SRPM (and in that SRPM will be the spec file itself), the rest of that function creates the parameters for the spec file to be expanded with (see _write_spec_file also for how requirements, epoch get determined). Building: aka, converting the SRPMs - RPMs 0. https://github.com/stackforge/anvil/blob/master/anvil/actions/build.py#L44 1. https://github.com/stackforge/anvil/blob/master/anvil/packaging/yum.py#L172 a. This is actually a simpler process since yum has a nice way to easily translate SRPMs-RPMs so all we do is use a make file with a parallel number of jobs to do that. So your hook could be at #2 if you wanted to not include the openvswitch logic in the spec file, u can hook in there to determine if say openvswitch is in this variable (which each instance/component that gets built should have populated, https://github.com/stackforge/anvil/blob/master/anvil/components/base.py#L30 and then easily switch on/off the spec file inclusion of the openvswitch packaging logic). I hope that makes sense, try printing out the subsystems variable (which comes in from https://github.com/stackforge/anvil/blob/master/conf/personas/in-a-box/basic.yaml#L37 , for example) . -Josh From: Sylvain Bauza sylvain.ba...@bull.netmailto:sylvain.ba...@bull.net Date: Wednesday, August 14, 2013 2:23 AM To: Joshua Harlow harlo...@yahoo-inc.commailto:harlo...@yahoo-inc.com Cc: OpenStack Development Mailing List openstack-dev@lists.openstack.orgmailto:openstack-dev@lists.openstack.org Subject: Re: [openstack-dev] [ANVIL] Missing openvswitch dependency for basic-neutron.yaml persona Thanks for the links, pretty useful. I do understand the process, but I have to admin I don't catch which Cheetah placeholder I would use for doing a big 'if' statement conditioning the package openstack-neutron-openvswitch on the core_plugin yaml option. As you said, this is not enough, if asked, openvswitch should also either be compiled or fetched from RDO. I filed a bug : https://bugs.launchpad.net/anvil/+bug/1212165 Anyway, I'm pretty much interested in doing the 1. you mentioned, I still need to understand things, tho. Could you be more precise on the way the spec files are populated ? Thanks, -Sylvain Le 13/08/2013 19:55, Joshua Harlow a écrit : Haha, no problem. Darn time differences. So some other useful links that I think will be helpful. - https://github.com/stackforge/anvil/blob/master/conf/templates/packaging/specs/openstack-neutron.spec This one is likely the biggest part of the issue, since it is the combination of all of neutron into 1 package (which has sub-packages). - One of those sub-packages is https://github.com/stackforge/anvil/blob/master/conf/templates/packaging/specs/openstack-neutron.spec#L274 This is pulling in the openvswitch part, that I think u don't want (at least not always want, it's wanted if neutron is going to use it, which under certain plugins it will). As you've seen it likely shouldn't be installing/needing that if https://github.com/stackforge/anvil/blob/master/anvil/components/configurators/neutron_plugins/linuxbridge.py is used. This should be coming from the following config (which will come from the yaml files) 'get_option' 'call': https://github.com/stackforge/anvil/blob/master/anvil/components/configurators/neutron.py#L49 So I think what can be done is a couple of things: 1. Don't include sub-packages that we don't want (the spec files are cheetahhttp://www.cheetahtemplate.org/ templates, so this can be done dynamically). 2. See if there is a way to make yum (or via yyoom) not pull in the dependencies for a sub-package when it won't be used (?) 3. Always build openvswitch (not as preferable) and include it (https://github.com/stackforge/anvil/blob/master/tools/build-openvswitch.sh) * I think the RDO repos might have some of these components. * http://openstack.redhat.com/Frequently_Asked_Questions#For_which_distributions_does_RDO_provide_packages.3F * This means we can just include the RDO repo rpm (like epel and use that openvswitch version there) instead of build your own. Hope some of this
Re: [openstack-dev] Skipping tests in tempest via config file
I put a nose.cfg with my excludes in the user's root and it works to run nosetests via the virtual environment like this: tempest/tools/./with_venv.sh nosetests I had to use the run_tests.sh script in tempest to create the virtual environment, but after that running tempest via nose within the venv wasn't a problem. Of course, I didn't want to duplicate the test runs when setting up the venv via run_tests.sh, so I created it with the -p option to only run pep8 after it was setup (I'm not aware of a way to tell it to not run any tests and simply setup the environment). Going back to the bug I opened last night for failures on py26, it's fixed with this patch: https://review.openstack.org/#/c/39346/ Thanks, MATT RIEDEMANN Advisory Software Engineer Cloud Solutions and OpenStack Development Phone: 1-507-253-7622 | Mobile: 1-507-990-1889 E-mail: mrie...@us.ibm.com 3605 Hwy 52 N Rochester, MN 55901-1407 United States From: Alexius Ludeman l...@lexinator.com To: OpenStack Development Mailing List openstack-dev@lists.openstack.org, Date: 08/14/2013 11:39 AM Subject:Re: [openstack-dev] Skipping tests in tempest via config file I am running tempest on OEL 6.3 (aka RHEL 6.3) and I had issues with python 2.6 and skipException[3], so now I'm using python 2.7 just for tempest. I also had difficulty with yum and python module dependency and made the transition to venv. This has reduced the yum dependency nightmare greatly. now that testr is default for tempest. testr does not appear to support --exclusion[1] or --stop[2]. I have a work around for --exclusion, by: testr list-tests | egrep -v regex-exclude-list unit-tests.txt testr --load-list unit-tests.txt I do not have a work around for --stop. [1]https://bugs.launchpad.net/testrepository/+bug/1208610 [2]https://bugs.launchpad.net/testrepository/+bug/1211926 [3]https://bugs.launchpad.net/tempest/+bug/1202815 On Tue, Aug 13, 2013 at 7:25 PM, Matt Riedemann mrie...@us.ibm.com wrote: I have the same issue. I run a subset of the tempest tests via nose on a RHEL 6.4 VM directly against the site-packages (not using virtualenv). I'm running on x86_64, ppc64 and s390x and have different issues on all of them (a mix of DB2 on x86_64 and MySQL on the others, and different nova/cinder drivers on each). What I had to do was just make a nose.cfg for each of them and throw that into ~/ for each run of the suite. The switch from nose to testr hasn't impacted me because I'm not using a venv. However, there was a change this week that broke me on python 2.6 and I opened this bug: https://bugs.launchpad.net/tempest/+bug/1212071 Thanks, MATT RIEDEMANN Advisory Software Engineer Cloud Solutions and OpenStack Development Phone: 1-507-253-7622 | Mobile: 1-507-990-1889 E-mail: mrie...@us.ibm.com 3605 Hwy 52 N Rochester, MN 55901-1407 United States From:Ian Wienand iwien...@redhat.com To:openstack-dev@lists.openstack.org, Date:08/13/2013 09:13 PM Subject:[openstack-dev] Skipping tests in tempest via config file Hi, I proposed a change to tempest that skips tests based on a config file directive [1]. Reviews were inconclusive and it was requested the idea be discussed more widely. Of course issues should go upstream first. However, sometimes test failures are triaged to a local/platform problem and it is preferable to keep everything else running by skipping the problematic tests while its being worked on. My perspective is one of running tempest in a mixed CI environment with RHEL, Fedora, etc. Python 2.6 on RHEL doesn't support testr (it doesn't do the setUpClass calls required by temptest) and nose upstream has some quirks that make it hard to work with the tempest test layout [2]. Having a common place in the temptest config to set these skips is more convienent than having to deal with the multiple testing environments. Another proposal is to have a separate JSON file of skipped tests. I don't feel strongly but it does seem like another config file. -i [1] https://review.openstack.org/#/c/39417/ [2] https://github.com/nose-devs/nose/pull/717 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev image/gifimage/gif___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
On Wed, 2013-08-14 at 12:35 -0300, Thierry Carrez wrote: Simo Sorce wrote: During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. Is ttx going to review any security implications ? The code does not mature just because is sit there untouched for more or less time. This is me wearing my vulnerability management hat on. The trick is that we (the VMT) have to support security issues for code that will be shipped in stable/havana. The most embarrassing security issues we had in the past were with code that didn't see a fair amount of time in master before we had to start supporting it. So for us there is a big difference between landing the KDS now and have it security-supported after one month of usage, and landing it in a few weeks and have it security-supported after 7 months of usage. After 7 months I'm pretty sure most of the embarrassing issues will be ironed out. I don't really want us to repeat the mistakes of the past where we shipped really new code in keystone that ended up not really usable, but which we still had to support security-wise due to our policy. By security implications, I mean that this is a domain (like, say, token expiration) where even basic bugs can easily create a vulnerability. We just don't have the bandwidth to ship an embargoed security advisory for every bug that will be found in the KDS one month from now. I understand and appreciate that, so are you saying you want to veto KDS introduction in Havana on this ground ? I would agree to this only if you can name individuals that are going to do a security review, otherwise I see no real reason to delay, as it will cost time to keep patches up to date, and I'd rather not do that if no one is lining up to do a security review. FWIW I did circulate the design for the security mechanism internally in Red Hat to some people with some expertise in crypto matters. Are you saying it won't have significantly less issues in 7 months just by the virtue of being landed in master and put into use in various projects ? Or that it was so thoroughly audited that my fears are unwarranted ? Bugs can always happen, and whether 7 month of being used in development makes a difference when it comes to security relevant bugs I can't say. I certainly am not going to claim my work flawless, I know better than that :) Simo. -- Simo Sorce * Red Hat, Inc * New York ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
On Wed, 2013-08-14 at 12:41 -0300, Thierry Carrez wrote: Dolph Mathews wrote: With regard to: https://blueprints.launchpad.net/keystone/+spec/key-distribution-server [...] Dolph: you don't mention Barbican at all, does that mean that the issue is settled and the KDS should live in keystone ? A side-benefit of landing early in Icehouse rather than late in Havana, IMHO, was that we could put everyone in the same room (and around the same beers) and get a single path forward. I'm a bit worried (that's with my release management hat on) that if everyone discovers that Barbican is the way to go for key distribution at the Hong-Kong summit, we would now have to deprecate the in-Keystone KDS over several releases just because it landed a few weeks too early. That said I haven't followed closely the latest discussions on this, so maybe this is not as much duplication of effort as I think ? For the Nth time KDS and Barbican do not do the same job, no more than Keystone auth paths and barbican do the same job. All three use crypto and 'keys' ... in completely different ways. Simo. -- Simo Sorce * Red Hat, Inc * New York ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
On Wed, 2013-08-14 at 14:06 -0300, Thierry Carrez wrote: Adam Young wrote: On 08/13/2013 06:20 PM, Dolph Mathews wrote: During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. I share the same concern, especially considering the API only recently went up for formal review [3], and the WIP implementation is still failing smokestack [4]. Since KDS is a security tightening in acase where there is no security at all, adding it in can only improve security. It's not really a question of more security or less security... It's about putting young sensitive code into a release, with the risk of having to issue a lot of security advisories for early bugs. I'm all for that code to land in the icehouse master branch as soon as it opens and that it gets put into good use by projects throughout the icehouse development cycle. I just think the benefits of waiting outweigh the benefits of landing it now. I explained why I prefer it to land in a few weeks rather than now... Can someone explain why they prefer the reverse ? Why does it have to be in havana ? Because it was painful top rebase due to the migrations code, however since Adam landed the code that splits migrations so that extensions can have their own separate code for that I think the burden will be substantially lower. If this is your final word on the matter I'll take notice that the work will be deferred till Icehouse and I will slightly demote its priority in my work queue. Simo. -- Simo Sorce * Red Hat, Inc * New York ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Migrating to testr parallel in tempest
I kind of high jacked another thread with my testr problems, but I want to reiterate it directly on this one as they are my pain points I had from the transition. testr does not appear to support --exclusion[1] or --stop[2]. I have a work around for --exclusion, by: testr list-tests | egrep -v regex-exclude-list unit-tests.txt testr --load-list unit-tests.txt I do not have a work around for --stop. [1]https://bugs.launchpad.net/testrepository/+bug/1208610 [2]https://bugs.launchpad.net/testrepository/+bug/1211926 On Tue, Aug 13, 2013 at 1:25 PM, Matthew Treinish mtrein...@kortar.orgwrote: Hi everyone, So for the past month or so I've been working on getting tempest to work stably with testr in parallel. As part of this you may have noticed the testr-full jobs that get run on the zuul check queue. I was using that job to debug some of the more obvious race conditions and stability issues with running tempest in parallel. After a bunch of fixes to tempest and finding some real bugs in some of the projects things seem to have smoothed out. So I pushed the testr-full run to the gate queue earlier today. I'll be keeping track of the success rate of this job vs the serial job and use this as the determining factor before we push this live to be the default for all tempest runs. So assuming that the success rate matches up well enough with serial job on the gate queue then I will push out the change that will migrate all the voting jobs to run in parallel hopefully either Friday afternoon or early next week. Also, if anyone has any input on what threshold they feel is good enough for this I'd welcome any input on that. For example, do we want to ensure a = 1:1 match for job success? Or would something like 90% as stable as the serial job be good enough considering the speed advantage. (The parallel runs take about half as much time as a full serial run, the parallel job normally finishes in ~25-30min) Since this affects almost every project I don't want to define this threshold without input from everyone. After there is some more data for the gate queue's parallel job I'll have some pretty graphite graphs that I can share comparing the success trends between the parallel and serial jobs. So at this point we're in the home stretch and I'm asking for everyone's help in getting this merged. So, if everyone who is reviewing and pushing commits could watch the results from these non-voting jobs and if things fail on the parallel job but not the serial job please investigate the failure and open a bug if necessary. If it turns out to be a bug in tempest please link it against this blueprint: https://blueprints.launchpad.net/tempest/+spec/speed-up-tempest so that I'll give it the attention it deserves. I'd hate to get this close to getting this merged and have a bit of racy code get merged at the last second and block us for another week or two. I feel that we need to get this in before the H3 rush starts up as it will help everyone get through the extra review load faster. Thanks, Matt Treinish ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [oslo] ack(), reject() and requeue() support in rpc ...
At Eric's request in https://review.openstack.org/#/c/41979/ I'm bringing this to the ML for feedback. Currently, oslo-common rpc behaviour is to always ack() a message no matter what. For billing purposes we can't afford to drop important notifications (like *.exists). We only want to ack() if no errors are raised by the consumer, otherwise we want to requeue the message. Now, once we introduce this functionality, we will also need to support .reject() semantics. The use-case we've seen for this is: 1. grab notification 2. write to disk 3. do some processing on that notification, which raises an exception. 4. the event is requeued and steps 2-3 repeat very quickly. Lots of duplicate records. In our case we've blown out our database. Since each notification has a unique message_id, it's easy to detect events we've seen before and .reject() them. There's a branch coming for that as well. This is really only a concern for the notification mechanism ... rpc should not need this, though I can think of lots of places it would be handy with orchestration and the scheduler. But, notifications are where the billing-related things live. By default, everything is the same, ack-on-error=True. But, for some consumers we need to support ack-on-error=False. Personally, I think this is must-have functionality. Look forward to your thoughts. -S ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ceilometer] SQL meter table schema improvements
- Original Message - From: Jay Pipes jaypi...@gmail.com To: OpenStack Development Mailing List openstack-dev@lists.openstack.org Sent: Wednesday, August 14, 2013 12:42:29 PM Subject: [openstack-dev] [Ceilometer] SQL meter table schema improvements Hi all, I submitted a bug report to improve the schema of the meter table to reduce duplication: https://bugs.launchpad.net/ceilometer/+bug/1211985 This would only affect the SQL storage driver. Interested in hearing thoughts on the above recommendations, which are: 1) Replace counter_type with an integer column counter_type_id that references a lookup table (counter_type) 2) Replace counter_unit with an integer column counter_unit_id that references a lookup table (counter_unit) 3) Replace counter_name with an integer column counter_id that references a lookup table (counter) Just those three changes would reduce the overall row size for records in the meter table by about 10%, by my calculations, which is a good improvement for a relatively painless set of modifications. Jay: Looks promising. Have you done any perf testing or do you have any numbers on how this affects performance other than the data size calculation? Thanks, -jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Migrating to testr parallel in tempest
- Original Message - From: Matthew Treinish mtrein...@kortar.org To: openstack-dev@lists.openstack.org Sent: Tuesday, August 13, 2013 4:25:14 PM Subject: [openstack-dev] Migrating to testr parallel in tempest Hi everyone, So for the past month or so I've been working on getting tempest to work stably with testr in parallel. As part of this you may have noticed the testr-full jobs that get run on the zuul check queue. I was using that job to debug some of the more obvious race conditions and stability issues with running tempest in parallel. After a bunch of fixes to tempest and finding some real bugs in some of the projects things seem to have smoothed out. So I pushed the testr-full run to the gate queue earlier today. I'll be keeping track of the success rate of this job vs the serial job and use this as the determining factor before we push this live to be the default for all tempest runs. So assuming that the success rate matches up well enough with serial job on the gate queue then I will push out the change that will migrate all the voting jobs to run in parallel hopefully either Friday afternoon or early next week. Also, if anyone has any input on what threshold they feel is good enough for this I'd welcome any input on that. For example, do we want to ensure a = 1:1 match for job success? Or would something like 90% as stable as the serial job be good enough considering the speed advantage. (The parallel runs take about half as much time as a full serial run, the parallel job normally finishes in ~25-30min) Since this affects almost every project I don't want to define this threshold without input from everyone. Nice work on the speedups! Regarding the stability... Having tests which fail due for stability reasons is concerning especially if it is as high as 10%. I personally get frustrated even at the 1% level. Lets see where the numbers fall. If we try it out we can always switch back if it is causing too many rechecks or failed gates right? If it does work great. If not I might prefer to see us keep a leaner set of gating Tempest tests so that runtime stays down in serial mode while we work bugs out of parallel mode. Dan After there is some more data for the gate queue's parallel job I'll have some pretty graphite graphs that I can share comparing the success trends between the parallel and serial jobs. So at this point we're in the home stretch and I'm asking for everyone's help in getting this merged. So, if everyone who is reviewing and pushing commits could watch the results from these non-voting jobs and if things fail on the parallel job but not the serial job please investigate the failure and open a bug if necessary. If it turns out to be a bug in tempest please link it against this blueprint: https://blueprints.launchpad.net/tempest/+spec/speed-up-tempest so that I'll give it the attention it deserves. I'd hate to get this close to getting this merged and have a bit of racy code get merged at the last second and block us for another week or two. I feel that we need to get this in before the H3 rush starts up as it will help everyone get through the extra review load faster. Thanks, Matt Treinish ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [neutron] Why is network and subnet modeled separately?
On 08/14/2013 04:31 PM, Jay Buffington wrote: network is layer 2 and subnet is layer 3. That's quite confusing. A common use case would be that one tenant creates a network with a 10.1.1.0/24 http://10.1.1.0/24 subnet. Another tenant wants to use that same network, so they create a new network and can also create the 10.1.1.0/24 http://10.1.1.0/24 subnet. Why not just call them all networks? -jay ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Skipping tests in tempest via config file
On 14 August 2013 14:10, Ian Wienand iwien...@redhat.com wrote: Hi, I proposed a change to tempest that skips tests based on a config file directive [1]. Reviews were inconclusive and it was requested the idea be discussed more widely. Of course issues should go upstream first. However, sometimes test failures are triaged to a local/platform problem and it is preferable to keep everything else running by skipping the problematic tests while its being worked on. There are arguments for and against having testr manage skipping. Testr definitely could do that, but should it? I suspect it's case by case. Skipping if you're on a small memory environment is really a test runner problem - you can probe for memory and skip. Skipping because Python2.6 doesn't run a bunch of tests likewise - it's internal to the test runner. Skipping because of a policy the test runner can't know about sounds like something testr can/should know about. -Rob -- Robert Collins rbtcoll...@hp.com Distinguished Technologist HP Converged Cloud ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [keystone] [oslo] postpone key distribution bp until icehouse?
Simo Sorce wrote: On Wed, 2013-08-14 at 12:35 -0300, Thierry Carrez wrote: Simo Sorce wrote: During today's project status meeting [1], the state of KDS was discussed [2]. To quote ttx directly: we've been bitten in the past with late security-sensitive stuff and I'm a bit worried to ship late code with such security implications as a KDS. Is ttx going to review any security implications ? The code does not mature just because is sit there untouched for more or less time. This is me wearing my vulnerability management hat on. The trick is that we (the VMT) have to support security issues for code that will be shipped in stable/havana. The most embarrassing security issues we had in the past were with code that didn't see a fair amount of time in master before we had to start supporting it. So for us there is a big difference between landing the KDS now and have it security-supported after one month of usage, and landing it in a few weeks and have it security-supported after 7 months of usage. After 7 months I'm pretty sure most of the embarrassing issues will be ironed out. I don't really want us to repeat the mistakes of the past where we shipped really new code in keystone that ended up not really usable, but which we still had to support security-wise due to our policy. By security implications, I mean that this is a domain (like, say, token expiration) where even basic bugs can easily create a vulnerability. We just don't have the bandwidth to ship an embargoed security advisory for every bug that will be found in the KDS one month from now. I understand and appreciate that, so are you saying you want to veto KDS introduction in Havana on this ground ? It's more of a trade-off: I want the benefits to exceed the drawbacks. Since I see this drawback, I'd like to understand the benefits so that we can collectively make the good trade-off... Does this really need to be in havana and why ? Or is it preferable to have it really early in icehouse ? Note that I can't really veto anything as long as the PTL wants it in :) Are you saying it won't have significantly less issues in 7 months just by the virtue of being landed in master and put into use in various projects ? Or that it was so thoroughly audited that my fears are unwarranted ? Bugs can always happen, and whether 7 month of being used in development makes a difference when it comes to security relevant bugs I can't say. I certainly am not going to claim my work flawless, I know better than that :) Damn, you escaped my trap :) -- Thierry Carrez (ttx) ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Migrating to testr parallel in tempest
On 15 August 2013 08:02, Alexius Ludeman l...@lexinator.com wrote: I kind of high jacked another thread with my testr problems, but I want to reiterate it directly on this one as they are my pain points I had from the transition. testr does not appear to support --exclusion[1] or --stop[2]. 1 is tricky to do well; 2 likewise but there is a trivial workaround in serial mode : tell the runner to stop early. I also replied in the other thread. -Rob -- Robert Collins rbtcoll...@hp.com Distinguished Technologist HP Converged Cloud ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] Migrating to testr parallel in tempest
On Wed, Aug 14, 2013 at 11:05:35AM -0500, Ben Nemec wrote: On 2013-08-13 16:39, Clark Boylan wrote: On Tue, Aug 13, 2013 at 1:25 PM, Matthew Treinish mtrein...@kortar.org wrote: Hi everyone, So for the past month or so I've been working on getting tempest to work stably with testr in parallel. As part of this you may have noticed the testr-full jobs that get run on the zuul check queue. I was using that job to debug some of the more obvious race conditions and stability issues with running tempest in parallel. After a bunch of fixes to tempest and finding some real bugs in some of the projects things seem to have smoothed out. So I pushed the testr-full run to the gate queue earlier today. I'll be keeping track of the success rate of this job vs the serial job and use this as the determining factor before we push this live to be the default for all tempest runs. So assuming that the success rate matches up well enough with serial job on the gate queue then I will push out the change that will migrate all the voting jobs to run in parallel hopefully either Friday afternoon or early next week. Also, if anyone has any input on what threshold they feel is good enough for this I'd welcome any input on that. For example, do we want to ensure a = 1:1 match for job success? Or would something like 90% as stable as the serial job be good enough considering the speed advantage. (The parallel runs take about half as much time as a full serial run, the parallel job normally finishes in ~25-30min) Since this affects almost every project I don't want to define this threshold without input from everyone. After there is some more data for the gate queue's parallel job I'll have some pretty graphite graphs that I can share comparing the success trends between the parallel and serial jobs. So at this point we're in the home stretch and I'm asking for everyone's help in getting this merged. So, if everyone who is reviewing and pushing commits could watch the results from these non-voting jobs and if things fail on the parallel job but not the serial job please investigate the failure and open a bug if necessary. If it turns out to be a bug in tempest please link it against this blueprint: https://blueprints.launchpad.net/tempest/+spec/speed-up-tempest so that I'll give it the attention it deserves. I'd hate to get this close to getting this merged and have a bit of racy code get merged at the last second and block us for another week or two. I feel that we need to get this in before the H3 rush starts up as it will help everyone get through the extra review load faster. Getting this in before the H3 rush would be very helpful. When we made the switch with Nova's unittests we fixed as many of the test bugs that we could find, merged the change to switch the test runner, then treated all failures as very high priority bugs that received immediate attention. Getting this in before H3 will give everyone a little more time to debug any potential new issues exposed by Jenkins or people running the tests locally. I think we should be bold here and merge this as soon as we have good numbers that indicate the trend is for these tests to pass. Graphite can give us the pass to fail ratios over time, as long as these trends are similar for both the old nosetest jobs and the new testr job I say we go for it. (Disclaimer: most of the projecst I work on are not affected by the tempest jobs; however, I am often called upon to help sort out issues in the gate). I'm inclined to agree. It's not as if we don't have transient failures now, and if we're looking at a 50% speedup in recheck/verify times then as long as the new version isn't significantly less stable it should be a net improvement. Of course, without hard numbers we're kind of discussing in a vacuum here. I also would like to get this in sooner rather than later and fix the bugs as they come in. But, I'm wary of doing this because there isn't a proven success history yet. No one likes gate resets, and I've only been running it on the gate queue for a day now. So here is the graphite graph that I'm using to watch parallel vs serial in the gate queue: https://tinyurl.com/pdfz93l On that graph the blue and yellow shows the number of jobs that succeeded grouped together in per hour buckets. (yellow being parallel and blue serial) Then the red line is showing failures, a horizontal bar means that there is no difference in the number of failures between serial and parallel. When it dips negative it is showing a failure in parallel that wasn't on serial a serial run at the same time. When it goes positive it showing a failure on serial that doesn't occur on parallel at the same time. But, because the serial runs take longer the failures happen at an offset. So if the plot shows parallel fails followed closely by a serial failure than that is probably on the same commit and
[openstack-dev] [State-Management] Agenda for tomorrows meeting at 2000 UTC
Hi all, The [state-management] project team holds a weekly meeting in #openstack-meeting on thursdays, 2000 UTC. The next meeting is tomorrow, Aug 14!!! As usual, everyone is welcome :-) Link: https://wiki.openstack.org/wiki/Meetings/StateManagement Taskflow: https://wiki.openstack.org/TaskFlow ## Agenda (30-60 mins): ## - Discuss any action items from last meeting. - Discuss ongoing status of the overall effort and any needed coordination. - Discuss about any new engine, block design concepts/reviews… - Talk about progress with regards to cinder/nova/ironic/heat/trove integration. - Discuss about any other potential new use-cases for said library. - Discuss about any other ideas, problems, issues, solutions, questions (and more). Any other topics are welcome :-) See you all soon! -Josh ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Glance] Blueprint proposal - Import / Export images with user properties
Thank's Mark, it's exactly this kind of feedback I was looking for. As you suggested, I'm going to make a single blueprint for OVF. Regards, Emilien Macchi # OpenStack Engineer // eNovance Inc. http://enovance.com // ✉ emil...@enovance.com ☎ +33 (0)1 49 70 99 80 // 10 rue de la Victoire 75009 Paris On 08/14/2013 07:39 PM, Mark Washenberger wrote: I think this could fit alongside a current blueprint we've discussed (https://blueprints.launchpad.net/glance/+spec/iso-image-metadata) that does similar things for metadata in isos. In general, I think the sane way to add a feature like this is as an optional container-format-specific plugin for import and export. Since the import and export features are still in pretty early stages of development (advanced on design though!), I don't expect such a feature would land until mid Icehouse, just fyi. Can you restructure these blueprints as a single bp feature to export/import metadata in ovf? On Wed, Aug 14, 2013 at 10:09 AM, Emilien Macchi emilien.mac...@enovance.com mailto:emilien.mac...@enovance.com wrote: Hi Mark, I was thinking at the OVF container format first since as far I know, it does support metadatas. Thank's, Emilien Macchi # OpenStack Engineer // eNovance Inc. http://enovance.com // ✉ emil...@enovance.com mailto:emil...@enovance.com ☎ +33 (0)1 49 70 99 80 tel:%2B33%20%280%291%2049%2070%2099%2080 // 10 rue de la Victoire 75009 Paris On 08/14/2013 06:34 PM, Mark Washenberger wrote: Lets dig into this a bit more so that I can understand it. Given that we have properties that we want to export with an image, where would those properties be stored? Somewhere in the image data itself? I believe some image formats support metadata, but I can't imagine all of them would. Is there a specific format you're thinking of using? On Wed, Aug 14, 2013 at 8:36 AM, Emilien Macchi emilien.mac...@enovance.com mailto:emilien.mac...@enovance.com wrote: Hi, I would like to discuss here about two blueprint proposal (maybe could I merge them into one if you prefer) : https://blueprints.launchpad.net/glance/+spec/api-v2-export-properties https://blueprints.launchpad.net/glance/+spec/api-v2-import-properties *Use case* : I would like to set specific properties to an image which could represent a signature, and useful for licensing requirements for example. To do that, I should be able to export an image with user properties included. Then, a user could reuse the exported image in the public cloud, and Glance will be aware about its properties. Obviously, we need the import / export feature. The idea here is to be able to identify an image after cloning or whatever with a property field. Of course, the user could break it in editing the image manually, but I consider he / she won't. Let me know if you have any thoughts and if the blueprint is valuable. Regards, -- Emilien Macchi # OpenStack Engineer // eNovance Inc. http://enovance.com // ✉ emil...@enovance.com mailto:emil...@enovance.com ☎ +33 (0)1 49 70 99 80 tel:%2B33%20%280%291%2049%2070%2099%2080 // 10 rue de la Victoire 75009 Paris ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org mailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org mailto:OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev signature.asc Description: OpenPGP digital signature ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [nova] Looking to get opinions on scheduler bug 1212478
For this bug: https://bugs.launchpad.net/nova/+bug/1212478 I can see the argument about not wanting the scheduler to keep trying to find a new host for the migration when the specific targeted host didn't work, but this seems like a design change so wanted to bring it up in the mailing list before trying to triage this. My initial thought is why have the RetryFilter configured for the scheduler if you don't want it to retry, but maybe you want to support both targeted and open (non-targeted) migrations. Thanks, MATT RIEDEMANN Advisory Software Engineer Cloud Solutions and OpenStack Development Phone: 1-507-253-7622 | Mobile: 1-507-990-1889 E-mail: mrie...@us.ibm.com 3605 Hwy 52 N Rochester, MN 55901-1407 United States image/gif___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [nova] Looking to get opinions on scheduler bug 1212478
Hi Matt-- After some additional investigation and sifting through code, I found that the RetryFilter and live migration are unrelated. I've posted some additional comments on there and invalidated this one. Thanks for the quick turnaround. Thanks, Joe Cropper 3605 Hwy 52 N Cloud Systems Software Development and x86 TCEM Rochester, MN 55901-1407 Phone: 507-253-1976 e-mail: jwcro...@us.ibm.com From: Matt Riedemann/Rochester/IBM@IBMUS To: OpenStack Development Mailing List openstack-dev@lists.openstack.org, Date: 08/14/2013 08:44 PM Subject:[openstack-dev] [nova] Looking to get opinions on scheduler bug 1212478 For this bug: https://bugs.launchpad.net/nova/+bug/1212478 I can see the argument about not wanting the scheduler to keep trying to find a new host for the migration when the specific targeted host didn't work, but this seems like a design change so wanted to bring it up in the mailing list before trying to triage this. My initial thought is why have the RetryFilter configured for the scheduler if you don't want it to retry, but maybe you want to support both targeted and open (non-targeted) migrations. Thanks, MATT RIEDEMANN Advisory Software Engineer Cloud Solutions and OpenStack Development Phone: 1-507-253-7622 | Mobile: 1-507-990-1889 E-mail: mrie...@us.ibm.com 3605 Hwy 52 N Rochester, MN 55901-1407 United States ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev image/gifimage/gif___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev