)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"0235ed98a2fe74f3696ab4f0cc3972b8a05d2c06","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":25,"id":"46cefbc6_eaed6161","updated":"2021-10-27 13:23:59.000000000","message":"recheck","commit_id":"4bee906e90eb9619e73bf5dfeaf8687e77bcce89"},{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"f72bce2935350d23f8effb2b9d325d4d52fed062","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":25,"id":"e9c3ca1f_ec8cf856","updated":"2021-10-28 01:03:49.000000000","message":"recheck","commit_id":"4bee906e90eb9619e73bf5dfeaf8687e77bcce89"},{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"e1119e9bb4c094ab5d2a42e85d979c41e2ae7047","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":25,"id":"fe33cbe3_a4234c62","updated":"2021-10-27 08:19:36.000000000","message":"recheck","commit_id":"4bee906e90eb9619e73bf5dfeaf8687e77bcce89"},{"author":{"_account_id":16137,"name":"Tobias Urdin","email":"tobias.urdin@binero.com","username":"tobasco"},"change_message_id":"b4a9cfd6d49a27164632562e46eb3c857684c36d","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"b5430359_0488023c","updated":"2021-11-17 08:15:24.000000000","message":"I was experimenting with that before for Keystone [1] using just plain Ruby and moving to the new Puppet Resource API for the type providers. I didn\u0027t actually see much of a performance improvement IIRC, I saved about 5-10 seconds per Puppet run, can probably just rerun that to get a better picture and read the resource report for how long it took in the Puppet log.\n\nNow the time might not be the important part there, I guess maintaining it would be the annoying part. I haven\u0027t read anything about Puppet deprecating the older type providers that might be a while forward but might just happend who knows, something to look out for.\n\n[1] https://review.opendev.org/q/topic:%22new-providers%22+(status:open%20OR%20status:merged) ","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"1b819c8328bf69a86cce0fffc6f8091d711deac2","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"c4acfe88_a12c695b","updated":"2021-11-29 14:17:56.000000000","message":"I\u0027ll move this ahead now since all related patches now have 2 +2s. There is one pending patch which introduces caching to optimize performance but that can be done as a follow-up.","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":14985,"name":"Alex Schultz","email":"aschultz@next-development.com","username":"mwhahaha"},"change_message_id":"ec6c53a7faa68399b044a35fcd1854eb5e1c5e4d","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"b54ee44f_ce2093e1","updated":"2021-11-15 15:36:54.000000000","message":"If we must, we must. But I do think it\u0027s important to raise the perf problems of openstackcli because if everything goes openstackcli only, it\u0027s probably going to cause problems","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":16137,"name":"Tobias Urdin","email":"tobias.urdin@binero.com","username":"tobasco"},"change_message_id":"0a4986e228fb18aa89fb5e2a76a0f29cdaa5765e","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"f1295f12_bb86ed89","updated":"2021-11-17 08:15:56.000000000","message":"TLDR: I agree this must be done, because there is no alternative, I\u0027ll leave you to +w it yourself after I\u0027ve reviewed everything.","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":14985,"name":"Alex Schultz","email":"aschultz@next-development.com","username":"mwhahaha"},"change_message_id":"c34a8a53543f4e7bd4bd37177a21b9b97a467cd6","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"bd1483fb_c0582175","updated":"2021-11-02 19:55:44.000000000","message":"btw openstack cli has a large perf penalty due to client loading.  We likely would need something like https://review.opendev.org/c/openstack/devstack/+/676016 to make it suck less","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"8626cc649de47f80c6ab81dd9457601f943d89ea","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":36,"id":"85d12f23_5e59268f","in_reply_to":"a9cc9ecf_07e42c25","updated":"2021-11-15 15:35:04.000000000","message":"Unfortunately neutron has decided to remove the neutron command during Z cycle[1], and I\u0027m afraid there is no alternative solutions atm unless we re-implement the whole client implementation(by Ruby ?)\n\n[1] https://review.opendev.org/c/openstack/python-neutronclient/+/793366\n\nAlthough there might be performance disadvantage with openstack cli, this migration has not yet caused any unreasonable degradation in CI which creates a few neutron resources so far.\n\nConsidering the clear removal plan, I tend to push this migration forward now to see any impact caused by the migration. This helps us investigate any alternative approach ealier in case we see any performance issue in real deployments.","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":14985,"name":"Alex Schultz","email":"aschultz@next-development.com","username":"mwhahaha"},"change_message_id":"a4229edf1198c3367ca88ec30786e5da9a27f6e4","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":36,"id":"6d96b293_44467e7c","in_reply_to":"b5430359_0488023c","updated":"2021-11-17 14:48:30.000000000","message":"The perf problems is likely dependent on the clients installed.  If the clients do stupid stuff in their loading like executing code to generate the cli params, it can slow stuff down.","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":9816,"name":"Takashi Kajinami","email":"kajinamit@oss.nttdata.com","username":"kajinamit"},"change_message_id":"98a7dd9be5eff3d684002553a85545dc769b3788","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":36,"id":"fe6ab18d_119c6070","in_reply_to":"bd1483fb_c0582175","updated":"2021-11-03 10:26:53.000000000","message":"Thanks. I\u0027ll follow that and see whether we can leverage the same. While implementing this change I found several points where we can introduce caching and I\u0027ll look into that later.\n\nAll of the implementation was created based on what we implemented by neutron cli by this series of patches and so far we don\u0027t see huge performance degradation in CI  so I tend to move forward this series now and do refactoring separately.\n\nJust FYI. Performance degradation I mentioned during internal meeting was not for this one but for https://review.opendev.org/c/openstack/puppet-keystone/+/806474 . It seems I made something wrong when I switched to OS_CLOUD or maybe usage of OS_CLOUD is slow...","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"},{"author":{"_account_id":14985,"name":"Alex Schultz","email":"aschultz@next-development.com","username":"mwhahaha"},"change_message_id":"b307773bf9e4b6373227e5af118f1d9f414e8ba3","unresolved":true,"context_lines":[],"source_content_type":"","patch_set":36,"id":"a9cc9ecf_07e42c25","in_reply_to":"fe6ab18d_119c6070","updated":"2021-11-03 13:55:57.000000000","message":"yea I know there\u0027s another one with OS_CLOUD, but openstackcli is another. I\u0027m concerned on the overall impact this will have due to openstackcli being awful depending on the plugins available.","commit_id":"241459ad567b48ad829e8f56d6064475278a79af"}]}