Re: [openstack-dev] [Ironic] Node groups and multi-node operations
On 27 January 2014 18:08, Joshua Harlow harlo...@yahoo-inc.com wrote: Thanks, guess this is entering the realm of scheduling group scheduling and how just the right level of information is needed to do efficient group scheduling in nova/ironic vs the new/upcoming gantt service. To me splitting it into N single requests isn't group scheduling but is just more of a batch processor to make things more parallel. To me it seems like gantt (or heat) or something else should know enough about the topology to identify where to schedule a request (or a group request) and then gantt/heat should pass enough location information to nova or ironic to let it know what was selected. Then nova or ironic can go about the dirty work of ensuring the instances were created reliably Of course it gets complicated when multiple resources are involved; but nobody said it was going to be easy ;) Right, it does get complex. One variation for instance - get a reservation from the scheduler which can schedule across network/block storage/compute and then follow that up with individual deployment requests that are tagged with the reservation. That way we don't need a mega-API that dispatches out to everything. Anyhow - I believe no matter what that this is future work and Ironic should be driven by the evolving design, rather than assuming anything about batch or not batch and changing in advance. -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] [Ironic] Node groups and multi-node operations
On Sat, Jan 25, 2014 at 7:11 AM, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-25 02:47:42 -0800: On 25 January 2014 19:42, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-24 18:48:41 -0800: However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? I can't find anything in Ironic that lets you group nodes by anything except chassis. It was not a serious discussion of how the problem would be solved, just a point that without some way to tie ironic nodes to compute-nodes I'd have to run multiple ironics. I don't understand the point. There is no tie between ironic nodes and compute nodes. Why do you want one? Because sans Ironic, compute-nodes still have physical characteristics that make grouping on them attractive for things like anti-affinity. I don't really want my HA instances not on the same compute node, I want them not in the same failure domain. It becomes a way for all OpenStack workloads to have more granularity than availability zone. Yes, and with Ironic, these same characteristics are desirable but are no longer properties of a nova-compute node; they're properties of the hardware which Ironic manages. In principle, the same (hypothetical) failure-domain-aware scheduling could be done if Ironic is exposing the same sort of group awareness, as long as the nova 'ironic driver is passing that information up to the scheduler in a sane way. In which case, Ironic would need to be representing such information, even if it's not acting on it, which I think is trivial for us to do. So if we have all of that modeled in compute-nodes, then when adding physical hardware to Ironic one just needs to have something to model the same relationship for each physical hardware node. We don't have to do it by linking hardware nodes to compute-nodes, but that would be doable for a first cut without much change to Ironic. You're trading failure-domain awareness for fault-tolerance in your control plane. by binding hardware to nova-compute. Ironic is designed explicitly to decouple the instances of Ironic (and Nova) within the control plane from the hardware it's managing. This is one of the main shortcomings of nova baremetal, and it doesn't seem like a worthy trade, even for a first approximation. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. I'm not sure I follow you at all. I'm suggesting that the scheduler have a new thing to filter on, and that compute nodes push their unique ID down into the Ironic driver so that while setting up nodes in Ironic one can assign them to a compute node. That doesn't sound massive and invasive. This is already being done *within* Ironic as nodes are mapped dynamically to ironic-conductor instances; the coordination for failover/takeover needs to be improved, but that's incremental at this point. Moving this mapping outside of Ironic is going to be messy and complicated, and breaks the abstraction layer. The API change may seem small, but it will massively overcomplicate Nova by duplicating all the functionality of Ironic-conductor in another layer of the stack. I think we're perhaps talking about different things - in the section you were answering, I thought he was talking about whether the API should offer operations on arbitrary sets of nodes at once, or whether each operation should be a separate API call vs what I now think you were talking about which was whether operations should be able to describe logical relations to other instances/nodes. Perhaps if we use the term 'batch' rather than 'group' to talk about the multiple-things-at-once aspect, and grouping to talk about the primarily scheduler related problems of affinity / anti affinity etc, we can avoid future confusion. Yes, thats a good point. I was talking about modeling failure domains only. Batching API requests seems like an entirely different thing. I was conflating these terms in that I was talking about grouping actions (batching) and groups of nodes (groups). That said, there are really three distinct topics here. Let's break groups down further: logical group for failure domains, and hardware group for hardware which is physically interdependent in such a way that changes to one node affect other node(s). Regards, Deva ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
Doesn't nova already have logic for creating N virtual machines (similar to a group) in the same request? I thought it did (maybe it doesn't anymore in the v3 API), creating N bare metal machines seems like it would comply to that api? Sent from my really tiny device... On Jan 22, 2014, at 4:50 PM, Devananda van der Veen devananda@gmail.com wrote: So, a conversation came again up today around whether or not Ironic will, in the future, support operations on groups of nodes. Some folks have expressed a desire for Ironic to expose operations on groups of nodes; others want Ironic to host the hardware-grouping data so that eg. Heat and Tuskar can make more intelligent group-aware decisions or represent the groups in a UI. Neither of these have an implementation in Ironic today... and we still need to implement a host of other things before we start on this. FWIW, this discussion is meant to stimulate thinking ahead to things we might address in Juno, and aligning development along the way. There's also some refactoring / code cleanup which is going on and worth mentioning because it touches the part of the code which this discussion impacts. For our developers, here is additional context: * our TaskManager class supports locking 1 node atomically, but both the driver API and our REST API only support operating on one node at a time. AFAIK, no where in the code do we actually pass a group of nodes. * for historical reasons, our driver API requires both a TaskManager and a Node object be passed to all methods. However, the TaskManager object contains a reference to the Node(s) which it has acquired, so the node parameter is redundant. * we've discussed cleaning this up, but I'd like to avoid refactoring the same interfaces again when we go to add group-awareness. I'll try to summarize the different axis-of-concern around which the discussion of node groups seem to converge... 1: physical vs. logical grouping - Some hardware is logically, but not strictly physically, grouped. Eg, 1U servers in the same rack. There is some grouping, such as failure domain, but operations on discrete nodes are discreet. This grouping should be modeled somewhere, and some times a user may wish to perform an operation on that group. Is a higher layer (tuskar, heat, etc) sufficient? I think so. - Some hardware _is_ physically grouped. Eg, high-density cartridges which share firmware state or a single management end-point, but are otherwise discrete computing devices. This grouping must be modeled somewhere, and certain operations can not be performed on one member without affecting all members. Things will break if each node is treated independently. 2: performance optimization - Some operations may be optimized if there is an awareness of concurrent identical operations. Eg, deploy the same image to lots of nodes using multicast or bittorrent. If Heat were to inform Ironic that this deploy is part of a group, the optimization would be deterministic. If Heat does not inform Ironic of this grouping, but Ironic infers it (eg, from timing of requests for similar actions) then optimization is possible but non-deterministic, and may be much harder to reason about or debug. 3: APIs - Higher layers of OpenStack (eg, Heat) are expected to orchestrate discrete resource units into a larger group operation. This is where the grouping happens today, but already results in inefficiencies when performing identical operations at scale. Ironic may be able to get around this by coalescing adjacent requests for the same operation, but this would be non-deterministic. - Moving group-awareness or group-operations into the lower layers (eg, Ironic) looks like it will require non-trivial changes to Heat and Nova, and, in my opinion, violates a layer-constraint that I would like to maintain. On the other hand, we could avoid the challenges around coalescing. This might be necessary to support physically-grouped hardware anyway, too. If Ironic coalesces requests, it could be done in either the ConductorManager layer or in the drivers themselves. The difference would be whether our internal driver API accepts one node or a set of nodes for each operation. It'll also impact our locking model. Both of these are implementation details that wouldn't affect other projects, but would affect our driver developers. Also, until Ironic models physically-grouped hardware relationships in some internal way, we're going to have difficulty supporting that class of hardware. Is that OK? What is the impact of not supporting such hardware? It seems, at least today, to be pretty minimal. Discussion is welcome. -Devananda ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
Excerpts from Devananda van der Veen's message of 2014-01-26 10:27:36 -0800: On Sat, Jan 25, 2014 at 7:11 AM, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-25 02:47:42 -0800: On 25 January 2014 19:42, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-24 18:48:41 -0800: However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? I can't find anything in Ironic that lets you group nodes by anything except chassis. It was not a serious discussion of how the problem would be solved, just a point that without some way to tie ironic nodes to compute-nodes I'd have to run multiple ironics. I don't understand the point. There is no tie between ironic nodes and compute nodes. Why do you want one? Because sans Ironic, compute-nodes still have physical characteristics that make grouping on them attractive for things like anti-affinity. I don't really want my HA instances not on the same compute node, I want them not in the same failure domain. It becomes a way for all OpenStack workloads to have more granularity than availability zone. Yes, and with Ironic, these same characteristics are desirable but are no longer properties of a nova-compute node; they're properties of the hardware which Ironic manages. I agree, but I don't see any of that reflected in Ironic's API. I see node CRUD, but not filtering or scheduling of any kind. In principle, the same (hypothetical) failure-domain-aware scheduling could be done if Ironic is exposing the same sort of group awareness, as long as the nova 'ironic driver is passing that information up to the scheduler in a sane way. In which case, Ironic would need to be representing such information, even if it's not acting on it, which I think is trivial for us to do. So if we have all of that modeled in compute-nodes, then when adding physical hardware to Ironic one just needs to have something to model the same relationship for each physical hardware node. We don't have to do it by linking hardware nodes to compute-nodes, but that would be doable for a first cut without much change to Ironic. You're trading failure-domain awareness for fault-tolerance in your control plane. by binding hardware to nova-compute. Ironic is designed explicitly to decouple the instances of Ironic (and Nova) within the control plane from the hardware it's managing. This is one of the main shortcomings of nova baremetal, and it doesn't seem like a worthy trade, even for a first approximation. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. I'm not sure I follow you at all. I'm suggesting that the scheduler have a new thing to filter on, and that compute nodes push their unique ID down into the Ironic driver so that while setting up nodes in Ironic one can assign them to a compute node. That doesn't sound massive and invasive. This is already being done *within* Ironic as nodes are mapped dynamically to ironic-conductor instances; the coordination for failover/takeover needs to be improved, but that's incremental at this point. Moving this mapping outside of Ironic is going to be messy and complicated, and breaks the abstraction layer. The API change may seem small, but it will massively overcomplicate Nova by duplicating all the functionality of Ironic-conductor in another layer of the stack. Can you point us to the design for this? I didn't really get that from browsing the code and docs, and I gave up trying to find a single architecture document after very little effort. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
On 27 January 2014 08:04, Joshua Harlow harlo...@yahoo-inc.com wrote: Doesn't nova already have logic for creating N virtual machines (similar to a group) in the same request? I thought it did (maybe it doesn't anymore in the v3 API), creating N bare metal machines seems like it would comply to that api? It does, but it splits it into N concurrent single server requests so that they get spread out amongst different nova-compute processes - getting you parallelisation: and the code for single server requests is sufficiently complex that having a rarely used path that preserves the batch seems undesirable to me. Besides which, as Ironic also dispatches work to many different backend workers, sending a batch to Ironic would just result in it having to split it out as well. -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] [Ironic] Node groups and multi-node operations
Thanks, guess this is entering the realm of scheduling group scheduling and how just the right level of information is needed to do efficient group scheduling in nova/ironic vs the new/upcoming gantt service. To me splitting it into N single requests isn't group scheduling but is just more of a batch processor to make things more parallel. To me it seems like gantt (or heat) or something else should know enough about the topology to identify where to schedule a request (or a group request) and then gantt/heat should pass enough location information to nova or ironic to let it know what was selected. Then nova or ironic can go about the dirty work of ensuring the instances were created reliably Of course it gets complicated when multiple resources are involved; but nobody said it was going to be easy ;) Sent from my really tiny device... On Jan 26, 2014, at 12:25 PM, Robert Collins robe...@robertcollins.net wrote: On 27 January 2014 08:04, Joshua Harlow harlo...@yahoo-inc.com wrote: Doesn't nova already have logic for creating N virtual machines (similar to a group) in the same request? I thought it did (maybe it doesn't anymore in the v3 API), creating N bare metal machines seems like it would comply to that api? It does, but it splits it into N concurrent single server requests so that they get spread out amongst different nova-compute processes - getting you parallelisation: and the code for single server requests is sufficiently complex that having a rarely used path that preserves the batch seems undesirable to me. Besides which, as Ironic also dispatches work to many different backend workers, sending a batch to Ironic would just result in it having to split it out as well. -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 ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
On 25 January 2014 19:42, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-24 18:48:41 -0800: However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? I can't find anything in Ironic that lets you group nodes by anything except chassis. It was not a serious discussion of how the problem would be solved, just a point that without some way to tie ironic nodes to compute-nodes I'd have to run multiple ironics. I don't understand the point. There is no tie between ironic nodes and compute nodes. Why do you want one? What makes you think this? Ironic runs in the same data centre as Nova... It it takes 2 Api calls to boot 1 physical machines is that really a performance problem? When other that first power on would you do that anyway? The API calls are meh. The image distribution and power fluctuations may not be. But there isn't a strong connection between API call and image distribution - e.g. (and this is my current favorite for 'when we get to optimising') a glance multicast service - Ironic would just add nodes to the relevant group as they are requested, and remove when they complete, and glance can take care of stopping the service when there are no members in the group. I actually think that the changes to Heat and Nova are trivial. Nova needs to have groups for compute nodes and the API needs to accept those groups. Heat needs to take advantage of them via the API. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. I'm not sure I follow you at all. I'm suggesting that the scheduler have a new thing to filter on, and that compute nodes push their unique ID down into the Ironic driver so that while setting up nodes in Ironic one can assign them to a compute node. That doesn't sound massive and invasive. I think we're perhaps talking about different things - in the section you were answering, I thought he was talking about whether the API should offer operations on arbitrary sets of nodes at once, or whether each operation should be a separate API call vs what I now think you were talking about which was whether operations should be able to describe logical relations to other instances/nodes. Perhaps if we use the term 'batch' rather than 'group' to talk about the multiple-things-at-once aspect, and grouping to talk about the primarily scheduler related problems of affinity / anti affinity etc, we can avoid future confusion. -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] [Ironic] Node groups and multi-node operations
Excerpts from Robert Collins's message of 2014-01-25 02:47:42 -0800: On 25 January 2014 19:42, Clint Byrum cl...@fewbar.com wrote: Excerpts from Robert Collins's message of 2014-01-24 18:48:41 -0800: However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? I can't find anything in Ironic that lets you group nodes by anything except chassis. It was not a serious discussion of how the problem would be solved, just a point that without some way to tie ironic nodes to compute-nodes I'd have to run multiple ironics. I don't understand the point. There is no tie between ironic nodes and compute nodes. Why do you want one? Because sans Ironic, compute-nodes still have physical characteristics that make grouping on them attractive for things like anti-affinity. I don't really want my HA instances not on the same compute node, I want them not in the same failure domain. It becomes a way for all OpenStack workloads to have more granularity than availability zone. So if we have all of that modeled in compute-nodes, then when adding physical hardware to Ironic one just needs to have something to model the same relationship for each physical hardware node. We don't have to do it by linking hardware nodes to compute-nodes, but that would be doable for a first cut without much change to Ironic. What makes you think this? Ironic runs in the same data centre as Nova... It it takes 2 Api calls to boot 1 physical machines is that really a performance problem? When other that first power on would you do that anyway? The API calls are meh. The image distribution and power fluctuations may not be. But there isn't a strong connection between API call and image distribution - e.g. (and this is my current favorite for 'when we get to optimising') a glance multicast service - Ironic would just add nodes to the relevant group as they are requested, and remove when they complete, and glance can take care of stopping the service when there are no members in the group. I think we agree here. Entirely. :) I actually think that the changes to Heat and Nova are trivial. Nova needs to have groups for compute nodes and the API needs to accept those groups. Heat needs to take advantage of them via the API. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. I'm not sure I follow you at all. I'm suggesting that the scheduler have a new thing to filter on, and that compute nodes push their unique ID down into the Ironic driver so that while setting up nodes in Ironic one can assign them to a compute node. That doesn't sound massive and invasive. I think we're perhaps talking about different things - in the section you were answering, I thought he was talking about whether the API should offer operations on arbitrary sets of nodes at once, or whether each operation should be a separate API call vs what I now think you were talking about which was whether operations should be able to describe logical relations to other instances/nodes. Perhaps if we use the term 'batch' rather than 'group' to talk about the multiple-things-at-once aspect, and grouping to talk about the primarily scheduler related problems of affinity / anti affinity etc, we can avoid future confusion. Yes, thats a good point. I was talking about modeling failure domains only. Batching API requests seems like an entirely different thing. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
Excerpts from Devananda van der Veen's message of 2014-01-22 16:44:01 -0800: 1: physical vs. logical grouping - Some hardware is logically, but not strictly physically, grouped. Eg, 1U servers in the same rack. There is some grouping, such as failure domain, but operations on discrete nodes are discreet. This grouping should be modeled somewhere, and some times a user may wish to perform an operation on that group. Is a higher layer (tuskar, heat, etc) sufficient? I think so. - Some hardware _is_ physically grouped. Eg, high-density cartridges which share firmware state or a single management end-point, but are otherwise discrete computing devices. This grouping must be modeled somewhere, and certain operations can not be performed on one member without affecting all members. Things will break if each node is treated independently. What Tuskar wants to do is layer workloads on top of logical and physical groupings. So it would pass to Nova Boot 4 machines with (flavor) and distinct(failure_domain_id) Now, this is not unique to baremetal. There are plenty of cloud workloads where one would like anti-affinity and other such things that will span more than a single compute node. Right now these are at a very coarse level which is availability zone. I think it is useful for Nova to be able to have a list of aspects for each compute node which are not hierarchical and isolate failure domains which matter for different work-loads. And with that, if we simply require at least one instance of nova-compute running for each set of aspects, Ironic does not have to model this data. However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. So perhaps if ironic can just model _a single_ logical grouping per node, it can defer any further distinctions up to Nova where it will benefit all workloads, not just Ironic. 2: performance optimization - Some operations may be optimized if there is an awareness of concurrent identical operations. Eg, deploy the same image to lots of nodes using multicast or bittorrent. If Heat were to inform Ironic that this deploy is part of a group, the optimization would be deterministic. If Heat does not inform Ironic of this grouping, but Ironic infers it (eg, from timing of requests for similar actions) then optimization is possible but non-deterministic, and may be much harder to reason about or debug. I'm wary of trying to get too deep on optimization this early. There are some blanket optimizations that you allude to here that I think will likely work o-k with even the most minimal of clues. 3: APIs - Higher layers of OpenStack (eg, Heat) are expected to orchestrate discrete resource units into a larger group operation. This is where the grouping happens today, but already results in inefficiencies when performing identical operations at scale. Ironic may be able to get around this by coalescing adjacent requests for the same operation, but this would be non-deterministic. Agreed, I think Ironic needs _some_ level of grouping to be efficient. - Moving group-awareness or group-operations into the lower layers (eg, Ironic) looks like it will require non-trivial changes to Heat and Nova, and, in my opinion, violates a layer-constraint that I would like to maintain. On the other hand, we could avoid the challenges around coalescing. This might be necessary to support physically-grouped hardware anyway, too. I actually think that the changes to Heat and Nova are trivial. Nova needs to have groups for compute nodes and the API needs to accept those groups. Heat needs to take advantage of them via the API. There is a non-trivial follow-on which is a wholistic scheduler which would further extend these groups into other physical resources like networks and block devices. These all feel like logical evolutions of the idea of making somewhat arbitrary and overlapping groups of compute nodes. If Ironic coalesces requests, it could be done in either the ConductorManager layer or in the drivers themselves. The difference would be whether our internal driver API accepts one node or a set of nodes for each operation. It'll also impact our locking model. Both of these are implementation details that wouldn't affect other projects, but would affect our driver developers. Also, until Ironic models physically-grouped hardware relationships in some internal way, we're going to have difficulty supporting that class of hardware. Is that OK? What is the impact of not supporting such hardware? It seems, at least today, to be pretty minimal. I don't know much about hardware like that. I think it should just be another grouping unless it affects the way Ironic talks to the hardware, at which point it probably belongs
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
On 25 Jan 2014 15:11, Clint Byrum cl...@fewbar.com wrote: Excerpts from Devananda van der Veen's message of 2014-01-22 16:44:01 -0800: What Tuskar wants to do is layer workloads on top of logical and physical groupings. So it would pass to Nova Boot 4 machines with (flavor) and distinct(failure_domain_id) Maybe. Maybe it would ask for a reservation and then ask for machines within that reservation Until it is unopened we are speculating :-) However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? So perhaps if ironic can just model _a single_ logical grouping per node, it can defer any further distinctions up to Nova where it will benefit all workloads, not just Ironic. Agreed with this. be deterministic. If Heat does not inform Ironic of this grouping, but Ironic infers it (eg, from timing of requests for similar actions) then optimization is possible but non-deterministic, and may be much harder to reason about or debug. I'm wary of trying to get too deep on optimization this early. There are some blanket optimizations that you allude to here that I think will likely work o-k with even the most minimal of clues. +1 premature optimisation and the root of all evil... 3: APIs r the same operation, but this would be non-deterministic. Agreed, I think Ironic needs _some_ level of grouping to be efficient. What makes you think this? Ironic runs in the same data centre as Nova... It it takes 2 Api calls to boot 1 physical machines is that really a performance problem? When other that first power on would you do that anyway? - Moving group-awareness or group-operations into the lower layers (eg, Ironic) looks like it will require non-trivial changes to Heat and Nova, and, in my opinion, violates a layer-constraint that I would like to maintain. On the other hand, we could avoid the challenges around coalescing. This might be necessary to support physically-grouped hardware anyway, too. I actually think that the changes to Heat and Nova are trivial. Nova needs to have groups for compute nodes and the API needs to accept those groups. Heat needs to take advantage of them via the API. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. There is a non-trivial follow-on which is a wholistic scheduler which would further extend these groups into other physical resources like networks and block devices. These all feel like logical evolutions of the idea of making somewhat arbitrary and overlapping groups of compute nodes. The holistic scheduler can also be a holistic reserver plus reservation aware scheduler -this avoids a lot of pain imo ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
On 23 Jan 2014 13:45, Devananda van der Veen devananda@gmail.com wrote: So, a conversation came again up today around whether or not Ironic will, in the future, support operations on groups of nodes. Some folks have expressed a desire for Ironic to expose operations on groups of nodes; others want Ironic to host the hardware-grouping data so that eg. Heat and Tuskar can make more intelligent group-aware decisions or represent the groups in a UI. Neither of these have an implementation in Ironic today... and we still need to implement a host of other things before we start on this. FWIW, this discussion is meant to stimulate thinking ahead to things we might address in Juno, and aligning development along the way. So I'm pretty thoroughly against this at this point in time. The rest of OpenStack has a single item at a time coding style ... Booting multiple instances is quickly transformed into n single instance boots. I think clear identification of services we need can take care of sharing workloads within ironic effectively. E.g. teach glance to multicast images and the story for doing many identical deploys at once becomes super simple. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
Re: [openstack-dev] [Ironic] Node groups and multi-node operations
Excerpts from Robert Collins's message of 2014-01-24 18:48:41 -0800: On 25 Jan 2014 15:11, Clint Byrum cl...@fewbar.com wrote: Excerpts from Devananda van der Veen's message of 2014-01-22 16:44:01 -0800: What Tuskar wants to do is layer workloads on top of logical and physical groupings. So it would pass to Nova Boot 4 machines with (flavor) and distinct(failure_domain_id) Maybe. Maybe it would ask for a reservation and then ask for machines within that reservation Until it is unopened we are speculating :-) Reservation is a better way to put it, I do agree with that. However, in looking at how Ironic works and interacts with Nova, it doesn't seem like there is any distinction of data per-compute-node inside Ironic. So for this to work, I'd have to run a whole bunch of ironic instances, one per compute node. That seems like something we don't want to do. Huh? I can't find anything in Ironic that lets you group nodes by anything except chassis. It was not a serious discussion of how the problem would be solved, just a point that without some way to tie ironic nodes to compute-nodes I'd have to run multiple ironics. So perhaps if ironic can just model _a single_ logical grouping per node, it can defer any further distinctions up to Nova where it will benefit all workloads, not just Ironic. Agreed with this. be deterministic. If Heat does not inform Ironic of this grouping, but Ironic infers it (eg, from timing of requests for similar actions) then optimization is possible but non-deterministic, and may be much harder to reason about or debug. I'm wary of trying to get too deep on optimization this early. There are some blanket optimizations that you allude to here that I think will likely work o-k with even the most minimal of clues. +1 premature optimisation and the root of all evil... 3: APIs r the same operation, but this would be non-deterministic. Agreed, I think Ironic needs _some_ level of grouping to be efficient. What makes you think this? Ironic runs in the same data centre as Nova... It it takes 2 Api calls to boot 1 physical machines is that really a performance problem? When other that first power on would you do that anyway? The API calls are meh. The image distribution and power fluctuations may not be. - Moving group-awareness or group-operations into the lower layers (eg, Ironic) looks like it will require non-trivial changes to Heat and Nova, and, in my opinion, violates a layer-constraint that I would like to maintain. On the other hand, we could avoid the challenges around coalescing. This might be necessary to support physically-grouped hardware anyway, too. I actually think that the changes to Heat and Nova are trivial. Nova needs to have groups for compute nodes and the API needs to accept those groups. Heat needs to take advantage of them via the API. The changes to Nova would be massive and invasive as they would be redefining the driver apiand all the logic around it. I'm not sure I follow you at all. I'm suggesting that the scheduler have a new thing to filter on, and that compute nodes push their unique ID down into the Ironic driver so that while setting up nodes in Ironic one can assign them to a compute node. That doesn't sound massive and invasive. There is a non-trivial follow-on which is a wholistic scheduler which would further extend these groups into other physical resources like networks and block devices. These all feel like logical evolutions of the idea of making somewhat arbitrary and overlapping groups of compute nodes. The holistic scheduler can also be a holistic reserver plus reservation aware scheduler -this avoids a lot of pain imo I think what I said still applies with that model, but it definitely becomes a lot more robust. ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
[openstack-dev] [Ironic] Node groups and multi-node operations
So, a conversation came again up today around whether or not Ironic will, in the future, support operations on groups of nodes. Some folks have expressed a desire for Ironic to expose operations on groups of nodes; others want Ironic to host the hardware-grouping data so that eg. Heat and Tuskar can make more intelligent group-aware decisions or represent the groups in a UI. Neither of these have an implementation in Ironic today... and we still need to implement a host of other things before we start on this. FWIW, this discussion is meant to stimulate thinking ahead to things we might address in Juno, and aligning development along the way. There's also some refactoring / code cleanup which is going on and worth mentioning because it touches the part of the code which this discussion impacts. For our developers, here is additional context: * our TaskManager class supports locking 1 node atomically, but both the driver API and our REST API only support operating on one node at a time. AFAIK, no where in the code do we actually pass a group of nodes. * for historical reasons, our driver API requires both a TaskManager and a Node object be passed to all methods. However, the TaskManager object contains a reference to the Node(s) which it has acquired, so the node parameter is redundant. * we've discussed cleaning this up, but I'd like to avoid refactoring the same interfaces again when we go to add group-awareness. I'll try to summarize the different axis-of-concern around which the discussion of node groups seem to converge... 1: physical vs. logical grouping - Some hardware is logically, but not strictly physically, grouped. Eg, 1U servers in the same rack. There is some grouping, such as failure domain, but operations on discrete nodes are discreet. This grouping should be modeled somewhere, and some times a user may wish to perform an operation on that group. Is a higher layer (tuskar, heat, etc) sufficient? I think so. - Some hardware _is_ physically grouped. Eg, high-density cartridges which share firmware state or a single management end-point, but are otherwise discrete computing devices. This grouping must be modeled somewhere, and certain operations can not be performed on one member without affecting all members. Things will break if each node is treated independently. 2: performance optimization - Some operations may be optimized if there is an awareness of concurrent identical operations. Eg, deploy the same image to lots of nodes using multicast or bittorrent. If Heat were to inform Ironic that this deploy is part of a group, the optimization would be deterministic. If Heat does not inform Ironic of this grouping, but Ironic infers it (eg, from timing of requests for similar actions) then optimization is possible but non-deterministic, and may be much harder to reason about or debug. 3: APIs - Higher layers of OpenStack (eg, Heat) are expected to orchestrate discrete resource units into a larger group operation. This is where the grouping happens today, but already results in inefficiencies when performing identical operations at scale. Ironic may be able to get around this by coalescing adjacent requests for the same operation, but this would be non-deterministic. - Moving group-awareness or group-operations into the lower layers (eg, Ironic) looks like it will require non-trivial changes to Heat and Nova, and, in my opinion, violates a layer-constraint that I would like to maintain. On the other hand, we could avoid the challenges around coalescing. This might be necessary to support physically-grouped hardware anyway, too. If Ironic coalesces requests, it could be done in either the ConductorManager layer or in the drivers themselves. The difference would be whether our internal driver API accepts one node or a set of nodes for each operation. It'll also impact our locking model. Both of these are implementation details that wouldn't affect other projects, but would affect our driver developers. Also, until Ironic models physically-grouped hardware relationships in some internal way, we're going to have difficulty supporting that class of hardware. Is that OK? What is the impact of not supporting such hardware? It seems, at least today, to be pretty minimal. Discussion is welcome. -Devananda ___ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev