)]}'
{"specs/rocky/cluster-supporting.rsa":[{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":11,"context_line":".."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":"Cluster supporting"},{"line_number":16,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":17,"context_line":""}],"source_content_type":"application/octet-stream","patch_set":2,"id":"9f6a8fd7_ff6133d4","line":14,"range":{"start_line":14,"start_character":18,"end_line":14,"end_character":43},"updated":"2018-05-03 09:43:38.000000000","message":"This extra \u003d should be removed","commit_id":"71907569b5ef5be8fc3cd3687f19e0c8443eb564"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":44,"context_line":"   b. We had accept network and az for instances in cluster[3], for mulit-azs"},{"line_number":45,"context_line":"env, we could meet error[4]."},{"line_number":46,"context_line":"   c. We also had multi-region support[5]. But for different regions, the nics"},{"line_number":47,"context_line":"are differnet, this could lose connection to other nodes in a cluter."},{"line_number":48,"context_line":"   d. Security group. Currently create sec_group for every node in a cluster,"},{"line_number":49,"context_line":"create one sec_group for a cluster seems more resonable."},{"line_number":50,"context_line":"   e. The limitation of node. For example, mongodb member node is limited"}],"source_content_type":"application/octet-stream","patch_set":2,"id":"9f6a8fd7_8af2326f","line":47,"range":{"start_line":47,"start_character":62,"end_line":47,"end_character":68},"updated":"2018-05-03 09:43:38.000000000","message":"cluster","commit_id":"71907569b5ef5be8fc3cd3687f19e0c8443eb564"}],"specs/rocky/cluster-supporting.rst":[{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":12,"context_line":""},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":"Cluster supporting"},{"line_number":16,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":17,"context_line":""},{"line_number":18,"context_line":".. If section numbers are desired, unindent this"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_889a1ac3","line":15,"range":{"start_line":15,"start_character":0,"end_line":15,"end_character":18},"updated":"2018-05-03 16:02:37.000000000","message":"I\u0027d prefer a more specific title here, though we could wait till we make the conclusion of what we\u0027re going to do exactly as the first step to improve cluster supporting.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":12,"context_line":""},{"line_number":13,"context_line":""},{"line_number":14,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":15,"context_line":"Cluster supporting"},{"line_number":16,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":17,"context_line":""},{"line_number":18,"context_line":".. If section numbers are desired, unindent this"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_8ca4640a","line":15,"range":{"start_line":15,"start_character":0,"end_line":15,"end_character":18},"in_reply_to":"5f7c97a3_889a1ac3","updated":"2018-05-04 08:24:05.000000000","message":"Agree, currently it\u0027s just a general topic. Wish to split into sub-steps in the further discussion.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":39,"context_line":"1. Cluster-create"},{"line_number":40,"context_line":""},{"line_number":41,"context_line":"* Before the node creation, trove validate the flavor, volume, nics[1][2]."},{"line_number":42,"context_line":"Since the consistency of these attributes are strictly required of each node"},{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_af76053e","line":42,"updated":"2018-05-03 09:43:38.000000000","message":"Quick format issue, if you want to use * at the beginning, then the lines after 1st line should have 2 spaces. Otherwise, the html generated by sphinx is not the same you expected. A reference of mine could be found here: \nhttps://review.openstack.org/#/c/557583/9/specs/rocky/adapt-to-file-injection-deprecation-in-nova.rst","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":39,"context_line":"1. Cluster-create"},{"line_number":40,"context_line":""},{"line_number":41,"context_line":"* Before the node creation, trove validate the flavor, volume, nics[1][2]."},{"line_number":42,"context_line":"Since the consistency of these attributes are strictly required of each node"},{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_e8d62ef2","line":42,"in_reply_to":"5f7c97a3_af76053e","updated":"2018-05-03 16:02:37.000000000","message":"Fan Zhang is right, and it\u0027s same for the ordered list, i.e use indentation to determine the content level.\n\nFor example, the correct syntax to arrange these 3 items with sublevel contents, we shoud do like:\n\n1. ITEM-1\n  * ITEM-1-SUB-1\n  * ITEM-1-SUB-2\n  ...\n\n2. ITEM-2\n\n  ITEM-2-SUB-1\n\n3. ITEM-3\n\n  ITEM-3-SUB-1\n\n(NOTE: indentations are necessary before \"ITEM-2-SUB-1\" and \"ITEM-3-SUB-1\" in the example)","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":39,"context_line":"1. Cluster-create"},{"line_number":40,"context_line":""},{"line_number":41,"context_line":"* Before the node creation, trove validate the flavor, volume, nics[1][2]."},{"line_number":42,"context_line":"Since the consistency of these attributes are strictly required of each node"},{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_ccd1fc59","line":42,"in_reply_to":"5f7c97a3_af76053e","updated":"2018-05-04 08:24:05.000000000","message":"thanks, will do a quick update to fix format issue.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":39,"context_line":"1. Cluster-create"},{"line_number":40,"context_line":""},{"line_number":41,"context_line":"* Before the node creation, trove validate the flavor, volume, nics[1][2]."},{"line_number":42,"context_line":"Since the consistency of these attributes are strictly required of each node"},{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_ecd63850","line":42,"in_reply_to":"5f7c97a3_e8d62ef2","updated":"2018-05-04 08:24:05.000000000","message":"thanks the example","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"* We had accept network and az for instances in cluster[3], for multi-azs"},{"line_number":47,"context_line":"environment, we could meet error[4]."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_a8991620","line":46,"range":{"start_line":46,"start_character":55,"end_line":46,"end_character":57},"updated":"2018-05-03 16:02:37.000000000","message":"It\u0027s better to use the gerrit changeset link as referrence, and then we could easily found out what was the change for, the discussions before it got merged. For this one, the gerrit change is:\n\nhttps://review.openstack.org/#/c/179443/\n\nand the related bug reported is:\n\nhttps://bugs.launchpad.net/trove/+bug/1447350\n\nIn the review history, some statements are still true:\n\n \u003e I updated the ticket with the details. nics and availability zones\n \u003e can be different for all the instances.\n \u003e \n \u003e In general nics parameter will be same, but people suggests that in\n \u003e certain situation user may want to provision certain set of\n \u003e instances in a separate network, so to provide that flexibility\n \u003e this option is made available on instance level for now.\n \u003e \n \u003e May be we should try to move some of these parameters like\n \u003e flavor/network on the cluster level because in most of the cases\n \u003e they will be same, and provide option to also specify it on the\n \u003e instance level to be able to customize few set of instances. if\n \u003e somebody needs to provision a 30 or 50 node cluster, where they\n \u003e don\u0027t need much customization they will be supplying redundant\n \u003e information for every instance. I think this change is  out of\n \u003e scope for this fix, but I think we will need that soon.\n\nor valuable to track the history of how the code base changed:\n\n \u003e Mongodb is little different. it provisions config server and query\n \u003e router for which one cannot specify any parameters like\n \u003e nics/flavors/az, as these two instance types are not explicitly\n \u003e mentioned in the cluster create call. I can still use the\n \u003e network/az information provided on the instance to provision rest\n \u003e of the cluster instances but then these special instance types will\n \u003e be provisioned without any network, which will make it\n \u003e inconsistent.\n\nand at last another spec which was not active and abondaned:\n\n \u003e I am  working on blueprint/spec, to address provisioning larger\n \u003e cluster where in most of these parameters (like flavor, volume,\n \u003e nics etc) will be supplied on the cluster level, which will also be\n \u003e used to address specific instances types which are provisioned\n \u003e behind the scenes.\n\nhttps://review.openstack.org/#/c/182859/\n\nBased all above changes or specs, I think multi-az support could be still valuable, so the problem is how we can improve the use interfaces (something like cluster level options maybe, or use cluster level options for cluster specific configurations and just allow only supply flavor/nics/etc for one instance and the other instances could reuse the same configuration) and more detailed documents about how to use different features in different deployments.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":43,"context_line":"in cluster, it seems unnecessary to pass them muiti-times and then do the"},{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"* We had accept network and az for instances in cluster[3], for multi-azs"},{"line_number":47,"context_line":"environment, we could meet error[4]."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_0cacd48a","line":46,"range":{"start_line":46,"start_character":55,"end_line":46,"end_character":57},"in_reply_to":"5f7c97a3_a8991620","updated":"2018-05-04 08:24:05.000000000","message":"Some of the references I have listed below(mongodb configure,routes, move some parameter to cluster level etc.).\nmulit-azs could discuss more, but I don\u0027t think multi-nics is a good choice, stability and performance of database is more import than flexibility for me, so maybe I prefer not support some \"certain situations\".","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":44,"context_line":"validation."},{"line_number":45,"context_line":""},{"line_number":46,"context_line":"* We had accept network and az for instances in cluster[3], for multi-azs"},{"line_number":47,"context_line":"environment, we could meet error[4]."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"},{"line_number":50,"context_line":"are differnet, this could lose connection to other nodes in a cluter."}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_23abe9d6","line":47,"range":{"start_line":47,"start_character":32,"end_line":47,"end_character":35},"updated":"2018-05-03 16:02:37.000000000","message":"The related bug reported is:\n\nhttps://bugs.launchpad.net/trove/+bug/1759102\n\nHowever the root cause is still not clear, and none of us has been do more investigation on the problem, :( And I personally don\u0027t think it\u0027s a problem caused by multi-az support.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":46,"context_line":"* We had accept network and az for instances in cluster[3], for multi-azs"},{"line_number":47,"context_line":"environment, we could meet error[4]."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"},{"line_number":50,"context_line":"are differnet, this could lose connection to other nodes in a cluter."},{"line_number":51,"context_line":""},{"line_number":52,"context_line":"* Currently create security group for each node in a cluster, create one"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_430e25e6","line":49,"range":{"start_line":49,"start_character":34,"end_line":49,"end_character":37},"updated":"2018-05-03 16:02:37.000000000","message":"Same situation here as the referrence [3].\n\nThe gerrit changeset is:\n\nhttps://review.openstack.org/#/c/382495/\n\nand the related spec is:\n\nhttps://review.openstack.org/#/c/294213/\n\nSo multi-region support is an special user case, we should make a decision about whether it\u0027s still valid to keep this feature or not.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"},{"line_number":50,"context_line":"are differnet, this could lose connection to other nodes in a cluter."},{"line_number":51,"context_line":""},{"line_number":52,"context_line":"* Currently create security group for each node in a cluster, create one"},{"line_number":53,"context_line":"security group for a cluster or a tenant seems more resonable."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* The limitation of node. For example, mongodb member node is limited"},{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_2f0d1590","line":53,"range":{"start_line":52,"start_character":62,"end_line":53,"end_character":61},"updated":"2018-05-03 09:43:38.000000000","message":"Couldn\u0027t agree more. We should definitely add the instances of one cluster into just one security group as they may have the same rules.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"},{"line_number":50,"context_line":"are differnet, this could lose connection to other nodes in a cluter."},{"line_number":51,"context_line":""},{"line_number":52,"context_line":"* Currently create security group for each node in a cluster, create one"},{"line_number":53,"context_line":"security group for a cluster or a tenant seems more resonable."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* The limitation of node. For example, mongodb member node is limited"},{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_aceb60ae","line":53,"range":{"start_line":52,"start_character":62,"end_line":53,"end_character":61},"in_reply_to":"5f7c97a3_0394ad43","updated":"2018-05-04 08:24:05.000000000","message":"Good point, didn\u0027t notice the effect of multi-regions to this. And currently I just keep these words unchange for future discuss.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":49,"context_line":"* We also had multi-region support[5]. But for different regions, the nics"},{"line_number":50,"context_line":"are differnet, this could lose connection to other nodes in a cluter."},{"line_number":51,"context_line":""},{"line_number":52,"context_line":"* Currently create security group for each node in a cluster, create one"},{"line_number":53,"context_line":"security group for a cluster or a tenant seems more resonable."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* The limitation of node. For example, mongodb member node is limited"},{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_0394ad43","line":53,"range":{"start_line":52,"start_character":62,"end_line":53,"end_character":61},"in_reply_to":"5f7c97a3_2f0d1590","updated":"2018-05-03 16:02:37.000000000","message":"This is true in most cases, but for the multi-region case, we need different SGs for instances in different regions(I searched about neutron SGs across regions, and it looks like this is not supported).","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":53,"context_line":"security group for a cluster or a tenant seems more resonable."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* The limitation of node. For example, mongodb member node is limited"},{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"},{"line_number":57,"context_line":"query router independently."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* Different mode of a datastore. For example, a user want to create a"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_6f622d19","line":56,"range":{"start_line":56,"start_character":3,"end_line":56,"end_character":11},"updated":"2018-05-03 09:43:38.000000000","message":"Yeah, actually, we have a customized modification here, adding a support for like 5\\7 members of a mongodb sharding cluster. But not elegant enough to push to upstream. It\u0027s definitely a gap needed to fix. As for the config server and query router things, IMHO, we need to have a strategy to specify these config items but not with flavor since it\u0027s very specific for mongodb, not for mysql, not for redis, even not for nova instances, if you know what I mean.  It seems something like configuration of datastore is one good choice here, I\u0027m just saying.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":53,"context_line":"security group for a cluster or a tenant seems more resonable."},{"line_number":54,"context_line":""},{"line_number":55,"context_line":"* The limitation of node. For example, mongodb member node is limited"},{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"},{"line_number":57,"context_line":"query router independently."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* Different mode of a datastore. For example, a user want to create a"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_a3bcd9bb","line":56,"range":{"start_line":56,"start_character":3,"end_line":56,"end_character":11},"in_reply_to":"5f7c97a3_6f622d19","updated":"2018-05-03 16:02:37.000000000","message":"Yes, I think this could be part of something like cluster level  options as I mentioned in the comment about multi-az support.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"},{"line_number":57,"context_line":"query router independently."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* Different mode of a datastore. For example, a user want to create a"},{"line_number":60,"context_line":"master-slave mysql cluster(with one or more slaves)by using a cluster-create"},{"line_number":61,"context_line":"API instead of calling instance-create API multi-times to achieve same goal."},{"line_number":62,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_aa1b5371","line":59,"range":{"start_line":59,"start_character":2,"end_line":59,"end_character":31},"updated":"2018-05-03 09:43:38.000000000","message":"I think the example you mentioned is because of that maybe the former developers treat mysql cluster and instance with slaves differently. It\u0027s OK to me though, but any improvement is welcomed. :)\n\nAs for multi-slaves for one instance, you can use parameter :\n```\n  --replica_count \u003ccount\u003e         Number of replicas to create (defaults to 1\n                                  if replica_of specified).\n```\nin command `trove create` of python-troveclient.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"},{"line_number":57,"context_line":"query router independently."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* Different mode of a datastore. For example, a user want to create a"},{"line_number":60,"context_line":"master-slave mysql cluster(with one or more slaves)by using a cluster-create"},{"line_number":61,"context_line":"API instead of calling instance-create API multi-times to achieve same goal."},{"line_number":62,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_8c34644c","line":59,"range":{"start_line":59,"start_character":2,"end_line":59,"end_character":31},"updated":"2018-05-04 08:24:05.000000000","message":"yes, the example maybe not suitable. The example is come from a true need from customers. For trove, the step of create a  one-master-multi-slaves mysql cluster is: 1. trove create 2. trove create --replica-count --replica-of. \nAnd, another mode trove can not do: master-master mysql mode(not galera)，because user don\u0027t want to promote a slave to master manually.\nWell, maybe it\u0027s a \"certain situation\", so we can focus on other features first.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":56,"context_line":"to three[6], and we can not specify the flavor, volume of config server and"},{"line_number":57,"context_line":"query router independently."},{"line_number":58,"context_line":""},{"line_number":59,"context_line":"* Different mode of a datastore. For example, a user want to create a"},{"line_number":60,"context_line":"master-slave mysql cluster(with one or more slaves)by using a cluster-create"},{"line_number":61,"context_line":"API instead of calling instance-create API multi-times to achieve same goal."},{"line_number":62,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_23ae892e","line":59,"range":{"start_line":59,"start_character":2,"end_line":59,"end_character":31},"in_reply_to":"5f7c97a3_aa1b5371","updated":"2018-05-03 16:02:37.000000000","message":"Agreed, the mysql example is not correct, replication and cluster are different features in Trove.\n\nBut cluster mode/type still make sense, for example, MongoDB replication has high availability feature, so it should be mapped to Trove cluster in fact(not Trove replication). This could also be true for other datastore backends(i.e. different cluster architectures are supported by the backend).","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":60,"context_line":"master-slave mysql cluster(with one or more slaves)by using a cluster-create"},{"line_number":61,"context_line":"API instead of calling instance-create API multi-times to achieve same goal."},{"line_number":62,"context_line":""},{"line_number":63,"context_line":"2. Cluster-action-grow"},{"line_number":64,"context_line":""},{"line_number":65,"context_line":"Currently, we pass flavor, volume, nics for cluster-grow, user could use a"},{"line_number":66,"context_line":"different value of there parameters with exists cluster node, maybe all node"},{"line_number":67,"context_line":"in a cluster has same flavor(volume, nics) is a better practices."},{"line_number":68,"context_line":""},{"line_number":69,"context_line":"3. Cluster-action-others"},{"line_number":70,"context_line":"Lack some **action** that instance has, such as cluster resize-volume, cluster"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_235c494a","line":67,"range":{"start_line":63,"start_character":0,"end_line":67,"end_character":65},"updated":"2018-05-03 16:02:37.000000000","message":"Different flavors/volumes/nics should be fine according the changes and specs mentioned in my previous comments, keep these  configurations same across cluster members should cover most user cases, however I think it\u0027s good to support some special cases and some people (or deployments) may need these features.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":66,"context_line":"different value of there parameters with exists cluster node, maybe all node"},{"line_number":67,"context_line":"in a cluster has same flavor(volume, nics) is a better practices."},{"line_number":68,"context_line":""},{"line_number":69,"context_line":"3. Cluster-action-others"},{"line_number":70,"context_line":"Lack some **action** that instance has, such as cluster resize-volume, cluster"},{"line_number":71,"context_line":"resize-flavor."},{"line_number":72,"context_line":""},{"line_number":73,"context_line":"Proposed Change"},{"line_number":74,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_03190c78","line":71,"range":{"start_line":69,"start_character":0,"end_line":71,"end_character":14},"updated":"2018-05-03 16:02:37.000000000","message":"As jian.song mentioned in the weekly meeting and the comment, the first step may be doing more investigation and some abstractions about cluster member restart. This could vary by different datastore backends, we can implement the feature one by one, and make this spec more specific and easier to be accomplished.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":75,"context_line":""},{"line_number":76,"context_line":"Here are what we intend to do:"},{"line_number":77,"context_line":""},{"line_number":78,"context_line":"* Treate some instance parameters as cluster parameters, like flavor, volume,"},{"line_number":79,"context_line":"nics, availability_zone, region_name. A request body could like::"},{"line_number":80,"context_line":""},{"line_number":81,"context_line":"    {\u0027cluster\u0027:"},{"line_number":82,"context_line":"        {\u0027volume\u0027:{\u0027size\u0027: 1, \u0027type\u0027: \u0027xxx\u0027},"},{"line_number":83,"context_line":"         \u0027nics\u0027: [{\u0027net-id\u0027: \u0027xxx\u0027}],"},{"line_number":84,"context_line":"         \u0027flavorRef\u0027: xxx,"},{"line_number":85,"context_line":"         \u0027datastore\u0027: {\u0027version\u0027: \u00273.6\u0027, \u0027type\u0027: \u0027mongodb\u0027},"},{"line_number":86,"context_line":"         \u0027name\u0027: \u0027test\u0027,"},{"line_number":87,"context_line":"         \u0027availability_zone\u0027: \u0027nova\u0027,"},{"line_number":88,"context_line":"         \u0027modules\u0027: xxx,"},{"line_number":89,"context_line":"         \u0027region_name\u0027: \u0027regionone\u0027,"},{"line_number":90,"context_line":"         \u0027configuration\u0027: xxx,"},{"line_number":91,"context_line":"         \u0027count\u0027: 3"},{"line_number":92,"context_line":"        }"},{"line_number":93,"context_line":"    }"},{"line_number":94,"context_line":""},{"line_number":95,"context_line":"The **count** means the number of data node in a cluster."},{"line_number":96,"context_line":""},{"line_number":97,"context_line":"* Take advantage of **extended_propertiesxx** [7]. For example, when create a"},{"line_number":98,"context_line":"mongodb cluster, we could pass extended_properties like::"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_63832006","line":95,"range":{"start_line":78,"start_character":0,"end_line":95,"end_character":57},"updated":"2018-05-03 16:02:37.000000000","message":"According to the above comments, this may need further investigation. The idea currently came to me is something like:\n{\u0027cluster\u0027:\n    {\u0027type\u0027: \u0027xxx\u0027,     #e.g. \u0027replication\u0027/\u0027sharding\u0027 for mongodb\n     \u0027members\u0027:\n         [{\u0027member_type\u0027: \u0027member\u0027,\n           \u0027config\u0027: \u0027symmetric\u0027,    # this could be the default, i.e. all members of this type use the same configurations\n            \u0027instances\u0027:\n                [{\u0027flavor\u0027: xxx, \u0027nics\u0027: xxx}, {}, {}],\n            \u0027instance_count\u0027: 3}]}    # for symmetric config type, we could only supply configurations for one instance, and specific \u0027instance_count\u0027; for asymmetric config type, the use should supply all instance configurations as a list just like the current implementation.\n\nHowever this kind of changes should be back compatible, we shouldn\u0027t break anything, otherwise we may need to consider a Trove v2 API.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":94,"context_line":""},{"line_number":95,"context_line":"The **count** means the number of data node in a cluster."},{"line_number":96,"context_line":""},{"line_number":97,"context_line":"* Take advantage of **extended_propertiesxx** [7]. For example, when create a"},{"line_number":98,"context_line":"mongodb cluster, we could pass extended_properties like::"},{"line_number":99,"context_line":""},{"line_number":100,"context_line":"    {\u0027extended_properties\u0027:"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_aac83374","line":97,"range":{"start_line":97,"start_character":22,"end_line":97,"end_character":43},"updated":"2018-05-03 09:43:38.000000000","message":"What\u0027s this? Which parameter of python-troveclient is it?\nI find this relevant patch: https://review.openstack.org/#/c/215775/","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":13157,"name":"Zhao Chao","email":"zhaochao1984@gmail.com","username":"zhaochao"},"change_message_id":"256f3a844ff32de166bec811122088c2f5505081","unresolved":false,"context_lines":[{"line_number":94,"context_line":""},{"line_number":95,"context_line":"The **count** means the number of data node in a cluster."},{"line_number":96,"context_line":""},{"line_number":97,"context_line":"* Take advantage of **extended_propertiesxx** [7]. For example, when create a"},{"line_number":98,"context_line":"mongodb cluster, we could pass extended_properties like::"},{"line_number":99,"context_line":""},{"line_number":100,"context_line":"    {\u0027extended_properties\u0027:"}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_83ebdcae","line":97,"range":{"start_line":97,"start_character":22,"end_line":97,"end_character":43},"in_reply_to":"5f7c97a3_aac83374","updated":"2018-05-03 16:02:37.000000000","message":"Same here, it\u0027s better use the gerrit url (Fan Zhang has mentioned this one) as referrence.\n\nUse \"extened_properties\" or new cluster level options still need further investigation and discussion.","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":25113,"name":"Fan Zhang","email":"zh.f@outlook.com","username":"fanzhang"},"change_message_id":"02901439b4e32452be8c2fb0a363f81d9441b0bb","unresolved":false,"context_lines":[{"line_number":112,"context_line":"    }"},{"line_number":113,"context_line":""},{"line_number":114,"context_line":"* Load the cluster info instead of pass them(flavor, nics, volume) during"},{"line_number":115,"context_line":"cluster-grow."},{"line_number":116,"context_line":""},{"line_number":117,"context_line":"* Consider a way to create security group for cluster or tentant instead"},{"line_number":118,"context_line":"of instance."}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_de8e94cf","line":115,"updated":"2018-05-03 09:43:38.000000000","message":"Load and do the verifying work, I guess?","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"},{"author":{"_account_id":22608,"name":"zhanggang","email":"zhanggang@cmss.chinamobile.com","username":"zhanggang"},"change_message_id":"688e8ac6bb7a7e7c9a6f1b2310d0de4e5f0bb48d","unresolved":false,"context_lines":[{"line_number":112,"context_line":"    }"},{"line_number":113,"context_line":""},{"line_number":114,"context_line":"* Load the cluster info instead of pass them(flavor, nics, volume) during"},{"line_number":115,"context_line":"cluster-grow."},{"line_number":116,"context_line":""},{"line_number":117,"context_line":"* Consider a way to create security group for cluster or tentant instead"},{"line_number":118,"context_line":"of instance."}],"source_content_type":"text/x-rst","patch_set":4,"id":"5f7c97a3_8ce9c481","line":115,"in_reply_to":"5f7c97a3_de8e94cf","updated":"2018-05-04 08:24:05.000000000","message":"not only do the verify, also use these parameters to create new cluster nodes()","commit_id":"213fa5a65ec8c1a108efc2c82029d294ff33c250"}]}
