Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
On 05/09/2015 04:23 PM, Krishnan Parthasarathi wrote: Oh nice, I might have missed the mails. Do you mind sharing the plan for 4.0? Any reason why you guys do not want to continue glusterd as translator model? I don't understand why we are using the translator model in the first place. I guess it was to reuse rpc code. You should be able to shed more light here. Even I am not sure :-). It was a translator by the time I got in. A quick google search with "glusterd 2.0 gluster-users", gave me this http://www.gluster.org/pipermail/gluster-users/2014-September/018639.html. Interestingly you asked us to consider AFR/NSR for distributed configuration management, which lead to http://www.gluster.org/pipermail/gluster-devel/2014-November/042944.html This proposal didn't go in the expected direction. I don't want to get into "why not use translators" now. We are currently heading in the direction visible in the above threads. If glusterd can't be a translator anymore, so be it. Kaushal's response gave the answers I was looking for. We should probably discuss it more once you guys come up with the interface CLI handling code needs to follow. I was thinking it would be great if you come up with a model where the handler code will be separate from the code of glusterd, which is what you guys seem to be targeting. Translator model is one way of achieving it, I personally love it on the FS side, that is why I was curious why it was not used. But any other way where the above requirements are met is welcome. Really excited to see what will come up :-). Pranith ___ Gluster-devel mailing list Gluster-devel@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
> Oh nice, I might have missed the mails. Do you mind sharing the plan for > 4.0? Any reason why you guys do not want to continue glusterd as > translator model? I don't understand why we are using the translator model in the first place. I guess it was to reuse rpc code. You should be able to shed more light here. A quick google search with "glusterd 2.0 gluster-users", gave me this http://www.gluster.org/pipermail/gluster-users/2014-September/018639.html. Interestingly you asked us to consider AFR/NSR for distributed configuration management, which lead to http://www.gluster.org/pipermail/gluster-devel/2014-November/042944.html This proposal didn't go in the expected direction. I don't want to get into "why not use translators" now. We are currently heading in the direction visible in the above threads. If glusterd can't be a translator anymore, so be it. ___ Gluster-devel mailing list Gluster-devel@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
On 05/09/2015 03:04 PM, Kaushal M wrote: Modularising GlusterD is something we plan to do. As of now, it's just that a plan. We don't yet have a design to achieve it yet. What Atin mentioned and what you've mentioned seem to be the same at a high level. The core of GlusterD will be a co-ordinating engine, which defines an interface for commands to use to do their work. The commands will each be a seperate module implementing this interface. Depending on how we implement, the actual names will be different. Yes, this is a nice approach. It would be nice if there is a clear demarcation as well for the code, so there won't be any dependency with merging dht changes vs say afr changes in cli. That is why I was suggesting xlator based solution. But other ways of doing it where there is clear demarcation is welcome as well. Would love to know more about the other approaches :-). Pranith On Sat, May 9, 2015 at 2:24 PM, Pranith Kumar Karampuri wrote: On 05/09/2015 02:21 PM, Atin Mukherjee wrote: On 05/09/2015 01:36 PM, Pranith Kumar Karampuri wrote: On 05/09/2015 11:08 AM, Krishnan Parthasarathi wrote: Ah! now I understood the confusion. I never said maintainer should fix all the bugs in tests. I am only saying that they maintain tests, just like we maintain code. Whether you personally work on it or not, you at least have an idea of what is the problem and what is the solution so someone can come and ask you and you know the status of it. Expectation is not to fix every test failure that comes maintainer's way by maintainer alone. But he/she would know about problem/solution because he/she at least reviews it and merges it. We want to make sure that the tests are in good quality as well just like we make sure code is of good quality. Core is a special case. We will handle it separately. Glusterd is also a 'special' case. As a glusterd maintainer, I am _not_ maintaining 's implementation. So, I don't 'know'/'understand' how it has been implemented and by extension I wouldn't be able to fix it (forget maintaining it :-) ). Given the no. of gluster commands, I won't be surprised if I didn't have an inkling on how worked ;-) I hope this encourages other contributors, i.e, any gluster (feature) contributor, to join Kaushal and me in maintaining glusterd. I understand the frustration kp :-). Human brain can only take so much. I think we are solving wrong problem by putting more people on the code. Why not break glusterd into small parts and distribute the load to different people? Did you guys plan anything for 4.0 for breaking glusterd? It is going to be a maintenance hell if we don't break it sooner. Glusterd does a lot of things: Lets see how we can break things up one thing at a time. I would love to spend some quality time thinking about this problem once I am done with ec work, but this is a rough idea I have for glusterd. 1) CLI handling: Glusterd-cli-xlator should act something like fuse in fs. It just gets the commands and passes it down, just like fuse gets the fops and passes it down. In glusterd process there should be snapshot.so, afr-cli.so, ec-cli.so, dht-cli.so loaded as management-xlators. Just like we have fops lets have mops (management operations). LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every time the top xlator in glusterd receives commands from cli, it converts the params into the arguments (req, op, dict etc) which are needed to carryout the cli. Now it winds the fop to all its children. One of the children is going to handle it locally, while the other child will send the cli to different glusterds that are in cluster. Second child of gluster-cli-xlator (give it a better name, but for now lets call it: mgmtcluster) will collate the responses and give the list of responses to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets call it local-handler) to collate the responses, i.e. logic for collating responses should also be in snapshot.so, afr-cli.so, dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, COMMIT-OP send response to CLI. 2) Volinfo should become more like inode_t in fs where each *-cli xlator can store their own ctx like snapshot-cli can store all snapshot related info for that volume in that context and afr can store afr-related info in the ctx. Volinfo data strcuture should have very minimal information. Maybe name, bricks etc. 3) Daemon handling: Daemon-manager xlator should have MOPS like START/STOP/INFO and this xlator should be accessible for all the -cli xlators which want to do their own management of the daemons. i.e. ec-cli/afr-cli should do self-heal-daemon handling. dht should do rebalance process handling etc. to give an example: while winding START mop it has to specify the daemon as "self-heal-daemon" and give enough info etc. 4) Peer handling: mgmtcluster(second child of top-xlator) should have MOPS like PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xl
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
On 05/09/2015 03:19 PM, Krishnan Parthasarathi wrote: Why not break glusterd into small parts and distribute the load to different people? Did you guys plan anything for 4.0 for breaking glusterd? It is going to be a maintenance hell if we don't break it sooner. Good idea. We have thought about it. Just re-architecting glusterd doesn't (and will not) solve the division of responsibility issue that is being discussed here. It's already difficult to maintain glusterd. I have already explained the reasons in the previous thread. I was thinking *-cli xlators could be maintained by the respective fs team itself. It is easier to maintain it this way because each of those xls can be put in xlators/cluster/afr/cli, xlators/cluster/dht/cli, etc. There will be clear demarcation of who owns what this way is my feeling. Even the tests can be organized to tests/afr-cli, tests/dht-cli etc etc. Glusterd does a lot of things: Lets see how we can break things up one thing at a time. I would love to spend some quality time thinking about this problem once I am done with ec work, but this is a rough idea I have for glusterd. 1) CLI handling: Glusterd-cli-xlator should act something like fuse in fs. It just gets the commands and passes it down, just like fuse gets the fops and passes it down. In glusterd process there should be snapshot.so, afr-cli.so, ec-cli.so, dht-cli.so loaded as management-xlators. Just like we have fops lets have mops (management operations). LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every time the top xlator in glusterd receives commands from cli, it converts the params into the arguments (req, op, dict etc) which are needed to carryout the cli. Now it winds the fop to all its children. One of the children is going to handle it locally, while the other child will send the cli to different glusterds that are in cluster. Second child of gluster-cli-xlator (give it a better name, but for now lets call it: mgmtcluster) will collate the responses and give the list of responses to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets call it local-handler) to collate the responses, i.e. logic for collating responses should also be in snapshot.so, afr-cli.so, dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, COMMIT-OP send response to CLI. 2) Volinfo should become more like inode_t in fs where each *-cli xlator can store their own ctx like snapshot-cli can store all snapshot related info for that volume in that context and afr can store afr-related info in the ctx. Volinfo data strcuture should have very minimal information. Maybe name, bricks etc. 3) Daemon handling: Daemon-manager xlator should have MOPS like START/STOP/INFO and this xlator should be accessible for all the -cli xlators which want to do their own management of the daemons. i.e. ec-cli/afr-cli should do self-heal-daemon handling. dht should do rebalance process handling etc. to give an example: while winding START mop it has to specify the daemon as "self-heal-daemon" and give enough info etc. 4) Peer handling: mgmtcluster(second child of top-xlator) should have MOPS like PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xlator is going to wind these operations based on the peer-cli-commands to this xlator. 5) volgen: top xlator is going to wind MOP called GET_NODE_LINKS, which takes the type of volfile (i.e. mount/nfs/shd/brick etc) on which each *-cli will construct its node(s), stuff options and tell the parent xl-name to which it needs to be linked to. Top xlator is going to just link the nodes to construct the graph and does graph_print to generate the volfile. I am pretty sure I forgot some more aspects of what glusterd does but you get the picture right? Break each aspect into different xlator and have MOPS to solve them. We have some initial ideas on how glusterd for 4.0 would look like. We won't be continuing with glusterd is also a translator model. The above model would work well only if we stuck with the stack of translators approach. Oh nice, I might have missed the mails. Do you mind sharing the plan for 4.0? Any reason why you guys do not want to continue glusterd as translator model? Pranith ___ Gluster-devel mailing list Gluster-devel@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
> Why not break glusterd into small parts and distribute the load to > different people? Did you guys plan anything for 4.0 for breaking glusterd? > It is going to be a maintenance hell if we don't break it sooner. Good idea. We have thought about it. Just re-architecting glusterd doesn't (and will not) solve the division of responsibility issue that is being discussed here. It's already difficult to maintain glusterd. I have already explained the reasons in the previous thread. > > Glusterd does a lot of things: Lets see how we can break things up one > thing at a time. I would love to spend some quality time thinking about > this problem once I am done with ec work, but this is a rough idea I > have for glusterd. > > 1) CLI handling: > Glusterd-cli-xlator should act something like fuse in fs. It just gets > the commands and passes it down, just like fuse gets the fops and passes > it down. In glusterd process there should be snapshot.so, afr-cli.so, > ec-cli.so, dht-cli.so loaded as management-xlators. > Just like we have fops lets have mops (management operations). > LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every > time the top xlator in glusterd receives commands from cli, it converts > the params into the arguments (req, op, dict etc) which are needed to > carryout the cli. Now it winds the fop to all its children. One of the > children is going to handle it locally, while the other child will send > the cli to different glusterds that are in cluster. Second child of > gluster-cli-xlator (give it a better name, but for now lets call it: > mgmtcluster) will collate the responses and give the list of responses > to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets > call it local-handler) to collate the responses, i.e. logic for > collating responses should also be in snapshot.so, afr-cli.so, > dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, > COMMIT-OP send response to CLI. > > 2) Volinfo should become more like inode_t in fs where each *-cli xlator > can store their own ctx like snapshot-cli can store all snapshot related > info for that volume in that context and afr can store afr-related info > in the ctx. Volinfo data strcuture should have very minimal information. > Maybe name, bricks etc. > > 3) Daemon handling: > Daemon-manager xlator should have MOPS like START/STOP/INFO and > this xlator should be accessible for all the -cli xlators which want to > do their own management of the daemons. i.e. ec-cli/afr-cli should do > self-heal-daemon handling. dht should do rebalance process handling etc. > to give an example: > while winding START mop it has to specify the daemon as > "self-heal-daemon" and give enough info etc. > > 4) Peer handling: > mgmtcluster(second child of top-xlator) should have MOPS like > PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xlator is going > to wind these operations based on the peer-cli-commands to this xlator. > > 5) volgen: > top xlator is going to wind MOP called GET_NODE_LINKS, which takes > the type of volfile (i.e. mount/nfs/shd/brick etc) on which each *-cli > will construct its node(s), stuff options and tell the parent xl-name to > which it needs to be linked to. Top xlator is going to just link the > nodes to construct the graph and does graph_print to generate the volfile. > > I am pretty sure I forgot some more aspects of what glusterd does but > you get the picture right? Break each aspect into different xlator and > have MOPS to solve them. We have some initial ideas on how glusterd for 4.0 would look like. We won't be continuing with glusterd is also a translator model. The above model would work well only if we stuck with the stack of translators approach. ___ Gluster-devel mailing list Gluster-devel@gluster.org http://www.gluster.org/mailman/listinfo/gluster-devel
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
Modularising GlusterD is something we plan to do. As of now, it's just that a plan. We don't yet have a design to achieve it yet. What Atin mentioned and what you've mentioned seem to be the same at a high level. The core of GlusterD will be a co-ordinating engine, which defines an interface for commands to use to do their work. The commands will each be a seperate module implementing this interface. Depending on how we implement, the actual names will be different. On Sat, May 9, 2015 at 2:24 PM, Pranith Kumar Karampuri wrote: > > On 05/09/2015 02:21 PM, Atin Mukherjee wrote: >> >> >> On 05/09/2015 01:36 PM, Pranith Kumar Karampuri wrote: >>> >>> On 05/09/2015 11:08 AM, Krishnan Parthasarathi wrote: > > Ah! now I understood the confusion. I never said maintainer should fix > all the bugs in tests. I am only saying that they maintain tests, just > like we maintain code. Whether you personally work on it or not, you at > least have an idea of what is the problem and what is the solution so > someone can come and ask you and you know the status of it. Expectation > is not to fix every test failure that comes maintainer's way by > maintainer alone. But he/she would know about problem/solution because > he/she at least reviews it and merges it. We want to make sure that the > tests are in good quality as well just like we make sure code is of > good > quality. Core is a special case. We will handle it separately. Glusterd is also a 'special' case. As a glusterd maintainer, I am _not_ maintaining 's implementation. So, I don't 'know'/'understand' how it has been implemented and by extension I wouldn't be able to fix it (forget maintaining it :-) ). Given the no. of gluster commands, I won't be surprised if I didn't have an inkling on how worked ;-) I hope this encourages other contributors, i.e, any gluster (feature) contributor, to join Kaushal and me in maintaining glusterd. >>> >>> I understand the frustration kp :-). Human brain can only take so much. >>> I think we are solving wrong problem by putting more people on the code. >>> Why not break glusterd into small parts and distribute the load to >>> different people? Did you guys plan anything for 4.0 for breaking >>> glusterd? >>> It is going to be a maintenance hell if we don't break it sooner. >>> >>> Glusterd does a lot of things: Lets see how we can break things up one >>> thing at a time. I would love to spend some quality time thinking about >>> this problem once I am done with ec work, but this is a rough idea I >>> have for glusterd. >>> >>> 1) CLI handling: >>> Glusterd-cli-xlator should act something like fuse in fs. It just gets >>> the commands and passes it down, just like fuse gets the fops and passes >>> it down. In glusterd process there should be snapshot.so, afr-cli.so, >>> ec-cli.so, dht-cli.so loaded as management-xlators. >>> Just like we have fops lets have mops (management operations). >>> LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every >>> time the top xlator in glusterd receives commands from cli, it converts >>> the params into the arguments (req, op, dict etc) which are needed to >>> carryout the cli. Now it winds the fop to all its children. One of the >>> children is going to handle it locally, while the other child will send >>> the cli to different glusterds that are in cluster. Second child of >>> gluster-cli-xlator (give it a better name, but for now lets call it: >>> mgmtcluster) will collate the responses and give the list of responses >>> to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets >>> call it local-handler) to collate the responses, i.e. logic for >>> collating responses should also be in snapshot.so, afr-cli.so, >>> dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, >>> COMMIT-OP send response to CLI. >>> >>> 2) Volinfo should become more like inode_t in fs where each *-cli xlator >>> can store their own ctx like snapshot-cli can store all snapshot related >>> info for that volume in that context and afr can store afr-related info >>> in the ctx. Volinfo data strcuture should have very minimal information. >>> Maybe name, bricks etc. >>> >>> 3) Daemon handling: >>> Daemon-manager xlator should have MOPS like START/STOP/INFO and >>> this xlator should be accessible for all the -cli xlators which want to >>> do their own management of the daemons. i.e. ec-cli/afr-cli should do >>> self-heal-daemon handling. dht should do rebalance process handling etc. >>> to give an example: >>> while winding START mop it has to specify the daemon as >>> "self-heal-daemon" and give enough info etc. >>> >>> 4) Peer handling: >>> mgmtcluster(second child of top-xlator) should have MOPS like >>> PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xlator is going >>> to wind these operations based on the peer-cli-commands to this
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
On 05/09/2015 02:21 PM, Atin Mukherjee wrote: On 05/09/2015 01:36 PM, Pranith Kumar Karampuri wrote: On 05/09/2015 11:08 AM, Krishnan Parthasarathi wrote: Ah! now I understood the confusion. I never said maintainer should fix all the bugs in tests. I am only saying that they maintain tests, just like we maintain code. Whether you personally work on it or not, you at least have an idea of what is the problem and what is the solution so someone can come and ask you and you know the status of it. Expectation is not to fix every test failure that comes maintainer's way by maintainer alone. But he/she would know about problem/solution because he/she at least reviews it and merges it. We want to make sure that the tests are in good quality as well just like we make sure code is of good quality. Core is a special case. We will handle it separately. Glusterd is also a 'special' case. As a glusterd maintainer, I am _not_ maintaining 's implementation. So, I don't 'know'/'understand' how it has been implemented and by extension I wouldn't be able to fix it (forget maintaining it :-) ). Given the no. of gluster commands, I won't be surprised if I didn't have an inkling on how worked ;-) I hope this encourages other contributors, i.e, any gluster (feature) contributor, to join Kaushal and me in maintaining glusterd. I understand the frustration kp :-). Human brain can only take so much. I think we are solving wrong problem by putting more people on the code. Why not break glusterd into small parts and distribute the load to different people? Did you guys plan anything for 4.0 for breaking glusterd? It is going to be a maintenance hell if we don't break it sooner. Glusterd does a lot of things: Lets see how we can break things up one thing at a time. I would love to spend some quality time thinking about this problem once I am done with ec work, but this is a rough idea I have for glusterd. 1) CLI handling: Glusterd-cli-xlator should act something like fuse in fs. It just gets the commands and passes it down, just like fuse gets the fops and passes it down. In glusterd process there should be snapshot.so, afr-cli.so, ec-cli.so, dht-cli.so loaded as management-xlators. Just like we have fops lets have mops (management operations). LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every time the top xlator in glusterd receives commands from cli, it converts the params into the arguments (req, op, dict etc) which are needed to carryout the cli. Now it winds the fop to all its children. One of the children is going to handle it locally, while the other child will send the cli to different glusterds that are in cluster. Second child of gluster-cli-xlator (give it a better name, but for now lets call it: mgmtcluster) will collate the responses and give the list of responses to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets call it local-handler) to collate the responses, i.e. logic for collating responses should also be in snapshot.so, afr-cli.so, dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, COMMIT-OP send response to CLI. 2) Volinfo should become more like inode_t in fs where each *-cli xlator can store their own ctx like snapshot-cli can store all snapshot related info for that volume in that context and afr can store afr-related info in the ctx. Volinfo data strcuture should have very minimal information. Maybe name, bricks etc. 3) Daemon handling: Daemon-manager xlator should have MOPS like START/STOP/INFO and this xlator should be accessible for all the -cli xlators which want to do their own management of the daemons. i.e. ec-cli/afr-cli should do self-heal-daemon handling. dht should do rebalance process handling etc. to give an example: while winding START mop it has to specify the daemon as "self-heal-daemon" and give enough info etc. 4) Peer handling: mgmtcluster(second child of top-xlator) should have MOPS like PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xlator is going to wind these operations based on the peer-cli-commands to this xlator. 5) volgen: top xlator is going to wind MOP called GET_NODE_LINKS, which takes the type of volfile (i.e. mount/nfs/shd/brick etc) on which each *-cli will construct its node(s), stuff options and tell the parent xl-name to which it needs to be linked to. Top xlator is going to just link the nodes to construct the graph and does graph_print to generate the volfile. I am pretty sure I forgot some more aspects of what glusterd does but you get the picture right? Break each aspect into different xlator and have MOPS to solve them. Sounds interesting but needs to be thought out in details. For 4.0,wWe do have a plan to make core glusterd algorithms work as a glusterd engine and other features will work have interfaces to connect to it. Your proposal looks another alternative. I would like to hear from the community about it. Whatever works to reduce the burden on you folks
Re: [Gluster-devel] break glusterd into small parts (Re: good job on fixing heavy hitters in spurious regressions)
On 05/09/2015 01:36 PM, Pranith Kumar Karampuri wrote: > > On 05/09/2015 11:08 AM, Krishnan Parthasarathi wrote: >>> Ah! now I understood the confusion. I never said maintainer should fix >>> all the bugs in tests. I am only saying that they maintain tests, just >>> like we maintain code. Whether you personally work on it or not, you at >>> least have an idea of what is the problem and what is the solution so >>> someone can come and ask you and you know the status of it. Expectation >>> is not to fix every test failure that comes maintainer's way by >>> maintainer alone. But he/she would know about problem/solution because >>> he/she at least reviews it and merges it. We want to make sure that the >>> tests are in good quality as well just like we make sure code is of good >>> quality. Core is a special case. We will handle it separately. >> Glusterd is also a 'special' case. As a glusterd maintainer, I am >> _not_ maintaining >> 's implementation. So, I >> don't >> 'know'/'understand' how it has been implemented and by extension I >> wouldn't be able >> to fix it (forget maintaining it :-) ). Given the no. of gluster >> commands, I won't be >> surprised if I didn't have an inkling on how >> worked ;-) >> I hope this encourages other contributors, i.e, any gluster (feature) >> contributor, >> to join Kaushal and me in maintaining glusterd. > I understand the frustration kp :-). Human brain can only take so much. > I think we are solving wrong problem by putting more people on the code. > Why not break glusterd into small parts and distribute the load to > different people? Did you guys plan anything for 4.0 for breaking glusterd? > It is going to be a maintenance hell if we don't break it sooner. > > Glusterd does a lot of things: Lets see how we can break things up one > thing at a time. I would love to spend some quality time thinking about > this problem once I am done with ec work, but this is a rough idea I > have for glusterd. > > 1) CLI handling: > Glusterd-cli-xlator should act something like fuse in fs. It just gets > the commands and passes it down, just like fuse gets the fops and passes > it down. In glusterd process there should be snapshot.so, afr-cli.so, > ec-cli.so, dht-cli.so loaded as management-xlators. > Just like we have fops lets have mops (management operations). > LOCK/STAGE/BRICK-OP/COMMIT-OP if there are more add them as well. Every > time the top xlator in glusterd receives commands from cli, it converts > the params into the arguments (req, op, dict etc) which are needed to > carryout the cli. Now it winds the fop to all its children. One of the > children is going to handle it locally, while the other child will send > the cli to different glusterds that are in cluster. Second child of > gluster-cli-xlator (give it a better name, but for now lets call it: > mgmtcluster) will collate the responses and give the list of responses > to glusterd-cli-xlator, it will call COLLATE mop on the first-child(lets > call it local-handler) to collate the responses, i.e. logic for > collating responses should also be in snapshot.so, afr-cli.so, > dht-cli.so etc etc. Once the top translator does LOCK, STAGE, BRICK-OP, > COMMIT-OP send response to CLI. > > 2) Volinfo should become more like inode_t in fs where each *-cli xlator > can store their own ctx like snapshot-cli can store all snapshot related > info for that volume in that context and afr can store afr-related info > in the ctx. Volinfo data strcuture should have very minimal information. > Maybe name, bricks etc. > > 3) Daemon handling: > Daemon-manager xlator should have MOPS like START/STOP/INFO and > this xlator should be accessible for all the -cli xlators which want to > do their own management of the daemons. i.e. ec-cli/afr-cli should do > self-heal-daemon handling. dht should do rebalance process handling etc. > to give an example: > while winding START mop it has to specify the daemon as > "self-heal-daemon" and give enough info etc. > > 4) Peer handling: > mgmtcluster(second child of top-xlator) should have MOPS like > PEER_ADD/PEER_DEL/PEER_UPDATE etc to do the needful. top xlator is going > to wind these operations based on the peer-cli-commands to this xlator. > > 5) volgen: > top xlator is going to wind MOP called GET_NODE_LINKS, which takes > the type of volfile (i.e. mount/nfs/shd/brick etc) on which each *-cli > will construct its node(s), stuff options and tell the parent xl-name to > which it needs to be linked to. Top xlator is going to just link the > nodes to construct the graph and does graph_print to generate the volfile. > > I am pretty sure I forgot some more aspects of what glusterd does but > you get the picture right? Break each aspect into different xlator and > have MOPS to solve them. Sounds interesting but needs to be thought out in details. For 4.0,wWe do have a plan to make core glusterd algorithms work as a glusterd engine and other features will work have interfaces to conne