Re: [openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
1. I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I'd rather stick with two actions: - Assembly (externally): fpb --build name - Installation (on master node): fuel --install-plugin name I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. I suggest to keep it simple for now as we have the issue mentioned by Evgeny: fuel client is supposed to work from other nodes, and we will need additional verification code in there. Also, to make it smooth, we will have to end up with a few more checks - like what if tarball is broken, what if we can't find install script in it, etc. I'd suggest to run it simple for 6.0, and then we will see how it's being used and what other limitations / issues we have around plugin installation and usage. We can consider to make this functionality as part of fuel client a bit later. Thanks, On Tue, Oct 21, 2014 at 6:57 PM, Vitaly Kramskikh vkramsk...@mirantis.com wrote: Hi, As for a separate section for plugins, I think we should not force it and leave this decision to a plugin developer, so he can create just a single checkbox or a section of the settings tab or a separate tab depending on plugin functionality. Plugins should be able to modify arbitrary release fields. For example, if Ceph was a plugin, it should be able to extend wizard config to add new options to Storage pane. If vCenter was a plugin, it should be able to set maximum amount of Compute nodes to 0. 2014-10-20 21:21 GMT+07:00 Evgeniy L e...@mirantis.com: Hi guys, *Romans' questions:* I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. What we can do is to try to determine where fuel-client is installed, if it's master node, we can perform installation, if it isn't master node, we can show user the message, that in the current version remote plugin installation is not supported. In the next versions if we implement plugin manager (which is separate service for plugins management) we will be able to do it remotely. How are we planning to distribute fuel plugin builder and its updates? Yes, as Mike mentioned our plan is to release it on PyPi which is python packages repository, so any developer will be able to run `pip install fpb` and get the tool. What happens if an error occurs during plugin installation? Plugins installation process is very simple, our plan is to have some kind of transaction, to make it atomic. 1. register plugin via API 2. copy the files In case of error on the 1st step, we can do nothing, in case of error on the 2nd step, remove files if there are any, and delete a plugin via rest api. And show user a message. What happens if an error occurs during plugin execution? In the first iteration we are going to interrupt deployment if there are any errors for plugin's tasks, also we are thinking how to improve it, for example we wanted to provide a special flag for each task, like fail_deploument_on_error, and only if it's true, we fail deployment in case of failed task. But it can be tricky to implement, it requires to change the current orchestrator/nailgun error handling logic. So, I'm not sure if we can implement this logic in the first release. Regarding to meaningful error messages, yes, we want to show the user, which plugin causes the error. Shall we consider a separate place in UI (tab) for plugins? +1 to Mike's answer When are we planning to focus on the 2 plugins which were identified as must-haves for 6.0? Cinder LBaaS For Cinder we are going to implement plugin which configures GlusterFS as cinder backend, so, if user has installed GlusterFS cluster, we can configure our cinder to work with it, I want to mention that we don't install GlusterFS nodes, we just configure cinder to work with user's GlusterFS cluster. Stanislaw B. already did some scripts which configures cinder to work with GlusterFS, so we are on testing stage. Regarding to LBaaS, Stanislaw B. did multinode implementation, ha implementation is tricky and requires some additional work, we are working on it. Nathan's questions: Looks like Mike answered UI related
Re: [openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
Hi Mike, I would like to add a bit more details about current implementation and how it can be done. *Implement installation as a scripts inside of tar ball:* Cons: * install script is really simple right now, but it will be much more complicated ** it requires to implement logic where we can ask user for login/password ** use some config, where we will be able to get endpoints, like where is keystone, nailgun ** validate that it's possible to install plugin on the current version of master ** handle error cases (to make installation process more atomic) * it will be impossible to deprecate the installation logic/method, because it's on the plugin's side and you cannot change a plugin which user downloaded some times ago, when we get plugin manager, we probably would like user to use plugin manager, instead of some scripts * plugin installation process is not so simple as it could be (untar, cd plugin, ./install) Pros: * plugin developer can change installation scripts (I'm not sure if it's a pros) *Add installation to fuel client:* Cons: * requires changes in fuel client, which are not good for fuel client by design (fuel client should be able to work remotely from user's machine), current implementation requires local operations on files, it will be changed in the future releases, so fuel-client will be able to do it via api, also we can determine if it's not master node by /etc/fuel/version.yaml and show the user a message which says that in the current version it's not possible to install the plugin remotely * plugin developer won't be able to change installation process (I'm not sure if it's a cons) Pros: * it's easier for user to install the plugin `fuel --install-plugin plugin_name-1.0.1.fpb' * all of the authentication logic already implemented in fuel client * fuel client uses config with endpoints which is generated by puppet * it will be easier to deprecate previous installation approach, we can just install new fuel client on the master which uses api Personally I like the second approach, and I think we should try to implement it, when we get time. Thanks, On Thu, Oct 23, 2014 at 3:02 PM, Mike Scherbakov mscherba...@mirantis.com wrote: 1. I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I'd rather stick with two actions: - Assembly (externally): fpb --build name - Installation (on master node): fuel --install-plugin name I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. I suggest to keep it simple for now as we have the issue mentioned by Evgeny: fuel client is supposed to work from other nodes, and we will need additional verification code in there. Also, to make it smooth, we will have to end up with a few more checks - like what if tarball is broken, what if we can't find install script in it, etc. I'd suggest to run it simple for 6.0, and then we will see how it's being used and what other limitations / issues we have around plugin installation and usage. We can consider to make this functionality as part of fuel client a bit later. Thanks, On Tue, Oct 21, 2014 at 6:57 PM, Vitaly Kramskikh vkramsk...@mirantis.com wrote: Hi, As for a separate section for plugins, I think we should not force it and leave this decision to a plugin developer, so he can create just a single checkbox or a section of the settings tab or a separate tab depending on plugin functionality. Plugins should be able to modify arbitrary release fields. For example, if Ceph was a plugin, it should be able to extend wizard config to add new options to Storage pane. If vCenter was a plugin, it should be able to set maximum amount of Compute nodes to 0. 2014-10-20 21:21 GMT+07:00 Evgeniy L e...@mirantis.com: Hi guys, *Romans' questions:* I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. What we can do is to try to determine where fuel-client is installed, if it's master node, we can perform installation, if it isn't master node, we can show user the message, that in the current version remote plugin installation is not supported. In the next versions if we implement plugin manager (which is
Re: [openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
Preventing plugin developers from implementing their own installer is a pro, not a con, you've already listed one reason in cons against install scripts inside plugin tarball: if we centralize plugin installation and management logic in fuel, we can change it once for all plugins and don't have to worry about old plugins using an obsolete installer. I think priorities here should be 1) easy of plugin development; and 2) ease of use. Pluggable architecture won't do us much good if we end up being the only ones being able to use it efficiently. Adding a little more complexity to fuelclient to allow moving a lot of fuel complexity from core to plugins is a good tradeoff. On Thu, Oct 23, 2014 at 8:32 AM, Evgeniy L e...@mirantis.com wrote: Hi Mike, I would like to add a bit more details about current implementation and how it can be done. Implement installation as a scripts inside of tar ball: Cons: * install script is really simple right now, but it will be much more complicated ** it requires to implement logic where we can ask user for login/password ** use some config, where we will be able to get endpoints, like where is keystone, nailgun ** validate that it's possible to install plugin on the current version of master ** handle error cases (to make installation process more atomic) * it will be impossible to deprecate the installation logic/method, because it's on the plugin's side and you cannot change a plugin which user downloaded some times ago, when we get plugin manager, we probably would like user to use plugin manager, instead of some scripts * plugin installation process is not so simple as it could be (untar, cd plugin, ./install) Pros: * plugin developer can change installation scripts (I'm not sure if it's a pros) Add installation to fuel client: Cons: * requires changes in fuel client, which are not good for fuel client by design (fuel client should be able to work remotely from user's machine), current implementation requires local operations on files, it will be changed in the future releases, so fuel-client will be able to do it via api, also we can determine if it's not master node by /etc/fuel/version.yaml and show the user a message which says that in the current version it's not possible to install the plugin remotely * plugin developer won't be able to change installation process (I'm not sure if it's a cons) Pros: * it's easier for user to install the plugin `fuel --install-plugin plugin_name-1.0.1.fpb' * all of the authentication logic already implemented in fuel client * fuel client uses config with endpoints which is generated by puppet * it will be easier to deprecate previous installation approach, we can just install new fuel client on the master which uses api Personally I like the second approach, and I think we should try to implement it, when we get time. Thanks, On Thu, Oct 23, 2014 at 3:02 PM, Mike Scherbakov mscherba...@mirantis.com wrote: I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I'd rather stick with two actions: Assembly (externally): fpb --build name Installation (on master node): fuel --install-plugin name I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. I suggest to keep it simple for now as we have the issue mentioned by Evgeny: fuel client is supposed to work from other nodes, and we will need additional verification code in there. Also, to make it smooth, we will have to end up with a few more checks - like what if tarball is broken, what if we can't find install script in it, etc. I'd suggest to run it simple for 6.0, and then we will see how it's being used and what other limitations / issues we have around plugin installation and usage. We can consider to make this functionality as part of fuel client a bit later. Thanks, On Tue, Oct 21, 2014 at 6:57 PM, Vitaly Kramskikh vkramsk...@mirantis.com wrote: Hi, As for a separate section for plugins, I think we should not force it and leave this decision to a plugin developer, so he can create just a single checkbox or a section of the settings tab or a separate tab depending on plugin functionality. Plugins should be able to modify arbitrary release fields. For example, if Ceph was a plugin, it should be able to extend wizard config to add new options to Storage pane. If vCenter was a plugin, it should be able to set maximum amount of Compute nodes to 0. 2014-10-20 21:21 GMT+07:00 Evgeniy L e...@mirantis.com: Hi guys, Romans' questions: I feel like we should not require user to unpack the plugin
Re: [openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
Hi, As for a separate section for plugins, I think we should not force it and leave this decision to a plugin developer, so he can create just a single checkbox or a section of the settings tab or a separate tab depending on plugin functionality. Plugins should be able to modify arbitrary release fields. For example, if Ceph was a plugin, it should be able to extend wizard config to add new options to Storage pane. If vCenter was a plugin, it should be able to set maximum amount of Compute nodes to 0. 2014-10-20 21:21 GMT+07:00 Evgeniy L e...@mirantis.com: Hi guys, *Romans' questions:* I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. What we can do is to try to determine where fuel-client is installed, if it's master node, we can perform installation, if it isn't master node, we can show user the message, that in the current version remote plugin installation is not supported. In the next versions if we implement plugin manager (which is separate service for plugins management) we will be able to do it remotely. How are we planning to distribute fuel plugin builder and its updates? Yes, as Mike mentioned our plan is to release it on PyPi which is python packages repository, so any developer will be able to run `pip install fpb` and get the tool. What happens if an error occurs during plugin installation? Plugins installation process is very simple, our plan is to have some kind of transaction, to make it atomic. 1. register plugin via API 2. copy the files In case of error on the 1st step, we can do nothing, in case of error on the 2nd step, remove files if there are any, and delete a plugin via rest api. And show user a message. What happens if an error occurs during plugin execution? In the first iteration we are going to interrupt deployment if there are any errors for plugin's tasks, also we are thinking how to improve it, for example we wanted to provide a special flag for each task, like fail_deploument_on_error, and only if it's true, we fail deployment in case of failed task. But it can be tricky to implement, it requires to change the current orchestrator/nailgun error handling logic. So, I'm not sure if we can implement this logic in the first release. Regarding to meaningful error messages, yes, we want to show the user, which plugin causes the error. Shall we consider a separate place in UI (tab) for plugins? +1 to Mike's answer When are we planning to focus on the 2 plugins which were identified as must-haves for 6.0? Cinder LBaaS For Cinder we are going to implement plugin which configures GlusterFS as cinder backend, so, if user has installed GlusterFS cluster, we can configure our cinder to work with it, I want to mention that we don't install GlusterFS nodes, we just configure cinder to work with user's GlusterFS cluster. Stanislaw B. already did some scripts which configures cinder to work with GlusterFS, so we are on testing stage. Regarding to LBaaS, Stanislaw B. did multinode implementation, ha implementation is tricky and requires some additional work, we are working on it. Nathan's questions: Looks like Mike answered UI related questions. Do we offer any kind of validation for settings on plug-ins? Or some way to for the developer to ensure that setting that cannot be default or computed get requested for the plug-in? Yes, each field can have regexp which is used during the validation. *Mike's questions:* One minor thing from me, which I forgot to mention during the demo: verbosity of fpb run. I understand it might sound like a bikeshedding now, but I believe if we develop it right from the very beginning, then we can save some time later. So I would suggest normal, short INFO output, and verbose one with --debug. Agree. Thanks for your feedback, On Sun, Oct 19, 2014 at 1:11 PM, Mike Scherbakov mscherba...@mirantis.com wrote: Hi all, I moved this conversation to openstack-dev to get a broader audience, since we started to discuss technical details. Raw notes from demo session: https://etherpad.openstack.org/p/cinder-neutron-plugins-second-demo. Let me start answering on a few questions below from Roman Nathan. How are we planning to distribute fuel plugin builder and its updates? Ideally, it should be available externally (outside of master node). I don't want us to repeat the same mistake as we did with Fuel client, which doesn't seem to be usable as an external dependency. The plan was to have Fuel
Re: [openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
Hi guys, *Romans' questions:* I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I like the idea of putting plugin installation functionality in fuel client, which is installed on master node. But in the current version plugin installation requires files operations on the master, as result we can have problems if user's fuel-client is installed on another env. What we can do is to try to determine where fuel-client is installed, if it's master node, we can perform installation, if it isn't master node, we can show user the message, that in the current version remote plugin installation is not supported. In the next versions if we implement plugin manager (which is separate service for plugins management) we will be able to do it remotely. How are we planning to distribute fuel plugin builder and its updates? Yes, as Mike mentioned our plan is to release it on PyPi which is python packages repository, so any developer will be able to run `pip install fpb` and get the tool. What happens if an error occurs during plugin installation? Plugins installation process is very simple, our plan is to have some kind of transaction, to make it atomic. 1. register plugin via API 2. copy the files In case of error on the 1st step, we can do nothing, in case of error on the 2nd step, remove files if there are any, and delete a plugin via rest api. And show user a message. What happens if an error occurs during plugin execution? In the first iteration we are going to interrupt deployment if there are any errors for plugin's tasks, also we are thinking how to improve it, for example we wanted to provide a special flag for each task, like fail_deploument_on_error, and only if it's true, we fail deployment in case of failed task. But it can be tricky to implement, it requires to change the current orchestrator/nailgun error handling logic. So, I'm not sure if we can implement this logic in the first release. Regarding to meaningful error messages, yes, we want to show the user, which plugin causes the error. Shall we consider a separate place in UI (tab) for plugins? +1 to Mike's answer When are we planning to focus on the 2 plugins which were identified as must-haves for 6.0? Cinder LBaaS For Cinder we are going to implement plugin which configures GlusterFS as cinder backend, so, if user has installed GlusterFS cluster, we can configure our cinder to work with it, I want to mention that we don't install GlusterFS nodes, we just configure cinder to work with user's GlusterFS cluster. Stanislaw B. already did some scripts which configures cinder to work with GlusterFS, so we are on testing stage. Regarding to LBaaS, Stanislaw B. did multinode implementation, ha implementation is tricky and requires some additional work, we are working on it. Nathan's questions: Looks like Mike answered UI related questions. Do we offer any kind of validation for settings on plug-ins? Or some way to for the developer to ensure that setting that cannot be default or computed get requested for the plug-in? Yes, each field can have regexp which is used during the validation. *Mike's questions:* One minor thing from me, which I forgot to mention during the demo: verbosity of fpb run. I understand it might sound like a bikeshedding now, but I believe if we develop it right from the very beginning, then we can save some time later. So I would suggest normal, short INFO output, and verbose one with --debug. Agree. Thanks for your feedback, On Sun, Oct 19, 2014 at 1:11 PM, Mike Scherbakov mscherba...@mirantis.com wrote: Hi all, I moved this conversation to openstack-dev to get a broader audience, since we started to discuss technical details. Raw notes from demo session: https://etherpad.openstack.org/p/cinder-neutron-plugins-second-demo. Let me start answering on a few questions below from Roman Nathan. How are we planning to distribute fuel plugin builder and its updates? Ideally, it should be available externally (outside of master node). I don't want us to repeat the same mistake as we did with Fuel client, which doesn't seem to be usable as an external dependency. The plan was to have Fuel Plugin Builder (fpb) on PyPI. Ideally it should be backward compatible with older Fuel release, i.e. when there is Fuel 7.0 out, you should be still able to create plugin for Fuel 6.0. If that it is going to be overcomplicated - I suggested to produce fpb for every Fuel release, and name it like fpb60, fpb61, fpb70, etc. Then it becomes easier to support and maintain plugin builders for certain versions of Fuel. Speaking about Fuel Client - there is no mistake. It's been discussed dozens of times, it's just lack of resources to make it on PyPI as well as to fix a few other things. I hope it could be done as part of efforts from [2]. -
[openstack-dev] [Fuel] Pluggable framework in Fuel: first prototype ready
Hi all, I moved this conversation to openstack-dev to get a broader audience, since we started to discuss technical details. Raw notes from demo session: https://etherpad.openstack.org/p/cinder-neutron-plugins-second-demo. Let me start answering on a few questions below from Roman Nathan. How are we planning to distribute fuel plugin builder and its updates? Ideally, it should be available externally (outside of master node). I don't want us to repeat the same mistake as we did with Fuel client, which doesn't seem to be usable as an external dependency. The plan was to have Fuel Plugin Builder (fpb) on PyPI. Ideally it should be backward compatible with older Fuel release, i.e. when there is Fuel 7.0 out, you should be still able to create plugin for Fuel 6.0. If that it is going to be overcomplicated - I suggested to produce fpb for every Fuel release, and name it like fpb60, fpb61, fpb70, etc. Then it becomes easier to support and maintain plugin builders for certain versions of Fuel. Speaking about Fuel Client - there is no mistake. It's been discussed dozens of times, it's just lack of resources to make it on PyPI as well as to fix a few other things. I hope it could be done as part of efforts from [2]. - Perhaps we have a separate settings tab just for Plug-Ins?For some complex plug-ins, they might require a dedicated tab. If we have too many tabs it could get messy. Shall we consider a separate place in UI (tab) for plugins? Settings tab seems to be overloaded. This is certainly under planning and discussion for future releases. See [1], for example. For 6.0, we agreed that we can just extend existing Settings tab with plugins-related fields. One minor thing from me, which I forgot to mention during the demo: verbosity of fpb run. I understand it might sound like a bikeshedding now, but I believe if we develop it right from the very beginning, then we can save some time later. So I would suggest normal, short INFO output, and verbose one with --debug. Thanks for feedback folks!!! [1] https://www.mail-archive.com/openstack-dev@lists.openstack.org/msg37196.html [2] https://www.mail-archive.com/openstack-dev@lists.openstack.org/msg37001.html -- Forwarded message -- From: Nathan Trueblood ntruebl...@mirantis.com Date: Sat, Oct 18, 2014 at 3:24 AM Subject: Re: plugins Agreed - I thought this initial PoC was great. A few initial thoughts about settings in the UI and plug-in in general: - Perhaps we have a separate settings tab just for Plug-Ins?For some complex plug-ins, they might require a dedicated tab. If we have too many tabs it could get messy. - It seems like we should consider how we handle the VMWare settings in light of plug-ins as well. Since with VMWare we have a lot of setting to configure and settings validation. - Do we offer any kind of validation for settings on plug-ins? Or some way to for the developer to ensure that setting that cannot be default or computed get requested for the plug-in? - We need to think carefully about both the plug-in developer experience (how hard to test, get error messages, etc) and the experience for the user who deploys the plug-in into an environment. -Nathan On Fri, Oct 17, 2014 at 4:13 PM, Roman Alekseenkov ralekseen...@mirantis.com wrote: I watched both videos (creating a file with the text from UI installing and starting a service). It looks pretty good!! Some initial feedback/questions: 1. I like the fact that fuel plugin builder appends version to the name and makes it fuel-awesome-plugin-1.2.3.tar. The approach is similar to Java/Maven and is a good one. 2. I feel like we should not require user to unpack the plugin before installing it. Moreover, we may chose to distribute plugins in our own format, which we may potentially change later. E.g. lbaas-v2.0.fp. I'd rather stick with two actions: - Assembly (externally): fpb --build name - Installation (on master node): fuel --install-plugin name 3. How are we planning to distribute fuel plugin builder and its updates? Ideally, it should be available externally (outside of master node). I don't want us to repeat the same mistake as we did with Fuel client, which doesn't seem to be usable as an external dependency. 4. How do we handle errors? - What happens if an error occurs during plugin installation? - What happens if an error occurs during plugin execution? Does it (should it?) fail the deployment? Will we show user an error message with the name of plugin that failed? 5. Shall we consider a separate place in UI (tab) for plugins? Settings tab seems to be overloaded. 6. When are we planning to focus on the 2 plugins which were identified as must-haves for 6.0? Cinder LBaaS Once again, great job guys! Thanks, Roman On Fri, Oct 17, 2014 at 9:32 AM, Mike Scherbakov mscherba...@mirantis.com wrote: Thanks, Evgeny,