| *** yizhihui has joined #openstack-smaug | 00:14 | |
| *** yinwei_computer has joined #openstack-smaug | 00:46 | |
| *** zhurong has joined #openstack-smaug | 01:05 | |
| *** yinwei_computer has quit IRC | 02:52 | |
| *** yinwei_computer has joined #openstack-smaug | 03:18 | |
| openstackgerrit | xiangxinyong proposed openstack/smaug: Fix Scheduled Operation Creation failed https://review.openstack.org/355709 | 03:59 |
|---|---|---|
| *** yinwei_computer has quit IRC | 04:27 | |
| *** yinwei_computer has joined #openstack-smaug | 06:27 | |
| *** yuval has joined #openstack-smaug | 06:29 | |
| *** yinwei_computer has quit IRC | 06:35 | |
| *** yinwei_computer has joined #openstack-smaug | 06:42 | |
| *** yinwei_computer has quit IRC | 06:45 | |
| *** yinwei_computer has joined #openstack-smaug | 06:51 | |
| *** yinwei_computer has quit IRC | 06:57 | |
| *** yinwei_computer has joined #openstack-smaug | 07:04 | |
| *** mingyu has joined #openstack-smaug | 07:05 | |
| *** yinwei_computer has quit IRC | 07:07 | |
| *** yinwei_computer has joined #openstack-smaug | 07:17 | |
| *** yinwei_computer has quit IRC | 07:20 | |
| *** yinwei_computer has joined #openstack-smaug | 07:27 | |
| *** yinwei_computer has quit IRC | 07:36 | |
| *** zengyingzhe has joined #openstack-smaug | 07:39 | |
| *** yinwei_computer has joined #openstack-smaug | 07:42 | |
| *** yinwei_computer has quit IRC | 07:44 | |
| *** yinwei_computer has joined #openstack-smaug | 07:52 | |
| *** yinwei_computer has quit IRC | 07:57 | |
| openstackgerrit | xiangxinyong proposed openstack/smaug-dashboard: Fix Trigger Creation Failed https://review.openstack.org/355771 | 07:59 |
| *** yinwei_computer has joined #openstack-smaug | 08:04 | |
| *** yinwei_computer has quit IRC | 08:07 | |
| *** yinwei_computer has joined #openstack-smaug | 08:29 | |
| *** yinwei_computer has quit IRC | 08:36 | |
| *** yinwei_computer has joined #openstack-smaug | 08:41 | |
| *** yamamoto has quit IRC | 08:44 | |
| *** yinwei_computer has quit IRC | 08:45 | |
| *** gampel has joined #openstack-smaug | 08:47 | |
| openstackgerrit | Yuval Brik proposed openstack/smaug: Fix Swift Bank Plugin Headers https://review.openstack.org/355427 | 08:51 |
| *** yinwei_computer has joined #openstack-smaug | 08:52 | |
| *** yinwei_computer has quit IRC | 08:57 | |
| *** gampel has quit IRC | 08:58 | |
| *** c00281451 is now known as zengchen | 09:00 | |
| xiangxinyong | should we start the meeting? | 09:02 |
| *** yinwei_computer has joined #openstack-smaug | 09:04 | |
| openstackgerrit | Merged openstack/smaug: Updated from global requirements https://review.openstack.org/350928 | 09:08 |
| *** yinwei_computer has quit IRC | 09:09 | |
| yuval | smile-luobin yinwei_computer are you coming to meeting? | 09:09 |
| yuval | zhonghua: ? | 09:09 |
| zhonghua | yuval:hi | 09:16 |
| yuval | zhonghua: ? | 09:16 |
| zhonghua | yuval: what's the matter? | 09:19 |
| openstackgerrit | Merged openstack/smaug-dashboard: Fix Trigger Creation Failed https://review.openstack.org/355771 | 09:22 |
| yuval | zhonghua: you said hey :) | 09:23 |
| *** yinwei_computer has joined #openstack-smaug | 09:26 | |
| zhonghua | yuval: ok :) | 09:27 |
| openstackgerrit | Merged openstack/smaug: Fix Scheduled Operation Creation failed https://review.openstack.org/355709 | 09:29 |
| *** yamamoto has joined #openstack-smaug | 09:31 | |
| *** yinwei_computer has quit IRC | 09:35 | |
| *** yinwei_computer has joined #openstack-smaug | 09:41 | |
| *** yamamoto has quit IRC | 09:41 | |
| openstackgerrit | Merged openstack/smaug-dashboard: create protection plan basic function https://review.openstack.org/328764 | 09:43 |
| openstackgerrit | Merged openstack/smaug-dashboard: create protection plan js function https://review.openstack.org/328765 | 09:44 |
| *** yinwei_computer has quit IRC | 09:45 | |
| *** yinwei_computer has joined #openstack-smaug | 09:45 | |
| openstackgerrit | Merged openstack/smaug-dashboard: implement protection plan detail page https://review.openstack.org/328766 | 09:46 |
| openstackgerrit | Merged openstack/smaug-dashboard: restore checkpoint basic function https://review.openstack.org/328807 | 09:46 |
| openstackgerrit | Merged openstack/smaug-dashboard: restore checkpoint js function https://review.openstack.org/328810 | 09:46 |
| openstackgerrit | Merged openstack/smaug-dashboard: implement checkpoint detail page https://review.openstack.org/328812 | 09:46 |
| *** yamamoto has joined #openstack-smaug | 09:46 | |
| *** yamamoto has quit IRC | 09:46 | |
| *** yamamoto has joined #openstack-smaug | 09:48 | |
| zhonghua | yuval xiangxinyong:do you have available time right now | 09:51 |
| openstackgerrit | Saggi Mizrahi proposed openstack/smaug: Correct logging translation markers https://review.openstack.org/352903 | 09:51 |
| xiangxinyong | yep | 09:51 |
| yuval | zhonghua: yes | 09:52 |
| yuval | zhonghua: not a lot of time, but yes | 09:52 |
| zhonghua | right now, we decide to store checkpoint into two places, one is bank anther one is db | 09:53 |
| zhonghua | is it possible to set bank as an optional one | 09:53 |
| zhonghua | that because in some use case that may not use bank-end | 09:53 |
| yuval | zhonghua: I thought we already said it is impossible. Some protection plugins store their data into the bank | 09:53 |
| zhonghua | yuval: I think it doesn't matter | 09:54 |
| yuval | zhonghua: what do you mean? | 09:54 |
| zhonghua | if anyone or any place want to use the bank, can also be configed | 09:55 |
| zhonghua | it's optional, not to remove it | 09:56 |
| yuval | bank is mandatory for a complete protection provider | 09:56 |
| chenying | yuval: There is a situation, smaug only be used in one website. The admin don't use our defaut plugins about iamge/network. They only want use local database to save checkpoint meatadata. They will not use bank as a backup backend. | 09:56 |
| yuval | we could have a 'null bank' | 09:56 |
| chenying | Yes | 09:56 |
| chenying | yuval: yes we mean it have a null back | 09:57 |
| yuval | problem is that bank is also used to keep checkpoint metadata | 09:57 |
| zhonghua | chenying: thanks for your description,that's the senario | 09:57 |
| chenying | yuval and we also need update the checkpoint api, can query the data form database | 09:57 |
| zhonghua | yuval: but right now, the checkpoint information had stored into the DB | 09:58 |
| yuval | wait wait | 09:58 |
| yuval | first, the checkpoint data and metadata are stored in the bank | 09:58 |
| zhonghua | https://review.openstack.org/#/c/351670/4/doc/source/specs/checkpoint-scheduled-clean.rst | 09:59 |
| yuval | the recent addition to store it to the database is not to change the original "checkpoint in bank" invariant | 09:59 |
| chenying | yuval: I plan to develop a database bank plugins before. only used save the checkpoint metadata to local database. | 10:00 |
| yuval | you could write a bank plugin to store the metadata in the database, but if you accidently use a protection plugin which protects the data in the bank you are in a problem | 10:00 |
| yuval | chenying: the bank plugin is not exposed to the content of the object, and whether it came from checkpoint metadata or data | 10:00 |
| *** zhurong has quit IRC | 10:01 | |
| yuval | chenying: you risk protecting tons of data to the database in case someone makes a mistake and combines this plugin with a protection plugin that protects into the bank | 10:01 |
| chenying | yuval: but now if xinyong have submit the bp about at the same time create the checkpoint meata data to database. only need implement a null back plugins. | 10:02 |
| zhonghua | yuval: I think that's not our problem, the deloyer should decide whether to use bank or not | 10:02 |
| yuval | zhonghua: what if the deployer doesn't use a bank but uses a protection plugin that backups an image to the bank? what happens? | 10:03 |
| openstackgerrit | Merged openstack/smaug: Fix Swift Bank Plugin Headers https://review.openstack.org/355427 | 10:03 |
| chenying | yuval: I know if the database backup being wrong used, it may be a porblom. | 10:03 |
| zhonghua | also, the adminstrator who config the protection plug-in must decide where to store the date | 10:03 |
| chenying | the database backup/ the database bank plugins | 10:03 |
| zhonghua | yuval: Karbor will reture the error | 10:04 |
| yuval | zhonghua: what error? karbor doesnt even know that the protection plugins is using the bank | 10:04 |
| zhonghua | yuval: it's a problem that how to get error from the plug-in | 10:05 |
| yuval | the checkpoint metadata in the database is not to be used instead of the bank | 10:06 |
| saggi | The checkpoint information in the operation engine can be user *only* by the operation engine. | 10:06 |
| zhonghua | yuval: If so. why to store the same information in two places? | 10:07 |
| saggi | The operation engine is an optional entity and doesn't know about any events outside of it | 10:07 |
| saggi | for example manual operation or operations done by other operation engine. | 10:07 |
| saggi | zhonghua: We do it now because we still don't have a good solution to query the checkpoint DB | 10:07 |
| saggi | it's one of the goals for v2 to allow the user to query the DB efficiently | 10:08 |
| saggi | by we are still no there | 10:08 |
| saggi | Also, this is what any other managing entity will do for itself since it might not work exclusively with Smaug | 10:08 |
| zhonghua | saggi: make sense. let's return our use case: some users do not want to deploy any bank-end, how to satisfy this requirement? | 10:09 |
| saggi | The information in the Operation Engine DB is also non authoritative. If a checkpoint is deleted manually it will still have an entry in the OperationEngine | 10:09 |
| chenying | yuval: I really have a senario from user that: admin don't want deploy a swift as a bank backend. they have their own backup backend to save data. the only need save the checkpoint metadata to database. | 10:09 |
| saggi | zhonghua: When we solve the bank indexing we will do it for single and multiple site | 10:09 |
| saggi | For single site the solution will obviously be much simpler | 10:10 |
| saggi | You can't have features that only work in the single site scenario. | 10:11 |
| zhonghua | saggi: yes, it's simple, but they think it's unreasonable to deploy an object storage system only for backup. | 10:11 |
| saggi | If they don't store chunks in the bank | 10:11 |
| saggi | We could have a DB based implemenation. | 10:11 |
| saggi | But you wouldn't be able to query the bank since the query solution needs to be generic. | 10:12 |
| zhonghua | saggi: it's difficult to implement the bank API via DB | 10:12 |
| *** yizhihui has quit IRC | 10:12 | |
| yuval | if the user wants to backup a volume they must have some kind of object storage for cinder backup backend | 10:12 |
| saggi | zhonghua: Not that much. And it's even easier to use NFS | 10:13 |
| *** yinwei_computer has quit IRC | 10:13 | |
| chenying | If they don't store chunks in the bank. Yes they have thrie own backup backend. will not use bank to save backup data. | 10:13 |
| zhonghua | saggi: more sense for NFS | 10:13 |
| xiangxinyong | lunch time? | 10:13 |
| saggi | xiangxinyong: yes :) | 10:14 |
| zhonghua | it's your lunch time , let us keep dicussing later. | 10:14 |
| xiangxinyong | sorry to interupt you:) | 10:14 |
| xiangxinyong | :) | 10:14 |
| saggi | zhonghua: The reason querying is so problematic is because requiring a globally replicatable document store is too much to ask. Object store is the minimum. | 10:14 |
| yuval | :) | 10:14 |
| saggi | DBs can't be geo replicated reliably. | 10:14 |
| saggi | At least for most implementations. | 10:15 |
| saggi | Object store\ File Systems are a much simpler use case | 10:15 |
| saggi | But querying is hard. | 10:15 |
| saggi | The current idea is to output events to searchlight (https://wiki.openstack.org/wiki/Searchlight) and have it handle the queries. | 10:17 |
| saggi | But this can't be made stable for V1 in the current timeline. | 10:18 |
| *** yinwei_computer has joined #openstack-smaug | 10:19 | |
| saggi | zhonghua, xiangxinyong ^^ | 10:19 |
| saggi | I see I'm the only one here so I'm going to eat. | 10:20 |
| *** mingyu has quit IRC | 10:21 | |
| *** mingyu has joined #openstack-smaug | 10:22 | |
| *** yamamoto has quit IRC | 10:25 | |
| *** yinwei_computer has quit IRC | 10:36 | |
| *** yinwei_computer has joined #openstack-smaug | 10:42 | |
| *** yinwei_computer has quit IRC | 10:45 | |
| *** yinwei_computer has joined #openstack-smaug | 10:52 | |
| *** yamamoto has joined #openstack-smaug | 11:03 | |
| *** yinwei_computer has quit IRC | 11:04 | |
| *** mingyu has quit IRC | 11:05 | |
| *** mingyu has joined #openstack-smaug | 11:06 | |
| *** mingyu has quit IRC | 11:10 | |
| xiangxinyong | :) | 11:12 |
| chenying | yuval saggi hi | 11:13 |
| yuval | chenying: hey | 11:13 |
| saggi | chenying: hi | 11:13 |
| chenying | wait. let me see what are you say before. | 11:14 |
| zhonghua | saggi: what can Searchlight do? make the index of main information and then search result by key msg? | 11:19 |
| saggi | zhonghua: You push the Metadata to it and it makes it searchable. | 11:21 |
| saggi | zhonghua: https://www.youtube.com/watch?v=2feC1njvZe0 | 11:22 |
| chenying | yuval saggi what's your oppion about the senario I have said. How to solve this requrement? | 11:22 |
| zhonghua | saggi: ok, but the problem I met is the user do not want to deploy so much dependent components. | 11:22 |
| saggi | That is how Cider is going to solve it's indexing problem. | 11:23 |
| zhonghua | saggi: from this point , I think use the searchlight or not is the same that can not resolve the problem | 11:23 |
| saggi | zhonghua: You are talking about having to install another service? | 11:26 |
| chenying | yuval: if the user wants to backup a volume they must have some kind of object storage for cinder backup backend. As I know, Block storege device is used in product solution as backup backend. They don't want intrduce a object storage to save checkpoint metadata, even intrduce another storeage like searchlight. | 11:27 |
| zhonghua | saggi: yeah, I met a problem that the user do not want to use Smaug without Swift. | 11:27 |
| saggi | chenying: We could make an FS base bank that you can use NFS or a local FS | 11:28 |
| saggi | or gluster | 11:28 |
| saggi | or whatever | 11:28 |
| saggi | it doesn't solve indexing | 11:28 |
| zhonghua | saggi: sorry, the user want to use Smaug without Swift | 11:28 |
| chenying | IMO checkpoint meata just is the record of one plan protection action. Why don't we save this data to database? | 11:29 |
| zhonghua | saggi: but, the fact is the swift is only bank-end implement right now, and it's a little difficult to develop a DB bank plugin. | 11:29 |
| saggi | chenying: Since it will prevent the Smaug bank from being easily geo-replicated. | 11:30 |
| saggi | which is one of the main requirements | 11:30 |
| saggi | We did have a thought about implementing something similar to searchlight with a DB internally. Where we send events to the DB when we make changes in the bank. But there are two issues with that: | 11:32 |
| saggi | 1. We are re-implementing searchlight. (though with an internal API so it's simpler to modify to our special needs) | 11:32 |
| saggi | 2. When you connect a new Bank the initial DB sync process could take a very long time. | 11:32 |
| saggi | If you use a DB you need to back up this DB. It creates a circular backup problem since backing up databases is complex. | 11:34 |
| saggi | This is why having something simple as the main storage is preferable. | 11:34 |
| saggi | zhonghua: ? | 11:37 |
| zhonghua | saggi: I think we can use DB A-A instead of DB backup. what's your opinion? | 11:37 |
| saggi | zhonghua: Active Active? | 11:38 |
| saggi | over WAN? | 11:38 |
| zhonghua | saggi: yeah, not necessary, the main senario is sigle-site back up and restore | 11:39 |
| saggi | zhonghua: You can't use do AA DB over WAN. It doesn't scale. If we ignore the multi-site use case who would solve it? How would you do DR? | 11:41 |
| zhonghua | I think it's unnecesary to over WAN. what's your suggestion? | 11:41 |
| saggi | zhonghua: eshed is asking about the Charter | 11:41 |
| zhonghua | saggi: I think there is some misunderstand on the sigle site backup and DR. | 11:42 |
| openstackgerrit | xiangxinyong proposed openstack/smaug: Rename Smaug DevStack to Karbor https://review.openstack.org/354469 | 11:44 |
| saggi | zhonghua: I think we can solve it with my suggestion. I'll send a BP for it. | 11:52 |
| zhonghua | saggi: use searchlight? | 11:53 |
| saggi | Internal indexing | 11:53 |
| zhonghua | saggi:thanks | 11:57 |
| *** gouthamr has joined #openstack-smaug | 12:32 | |
| xiangxinyong | saggi: could you review this patch? https://review.openstack.org/#/c/351670/ | 12:32 |
| xiangxinyong | thanks | 12:32 |
| *** yamamoto has quit IRC | 12:33 | |
| *** yamamoto has joined #openstack-smaug | 12:39 | |
| *** yamamoto has quit IRC | 12:43 | |
| saggi | xiangxinyong: sure | 12:45 |
| saggi | xiangxinyong: +2, just get another core to sign off on it | 12:46 |
| openstackgerrit | Merged openstack/smaug: Support checkpoint metadata https://review.openstack.org/350121 | 12:50 |
| *** yamamoto has joined #openstack-smaug | 13:07 | |
| *** yamamoto has quit IRC | 13:07 | |
| *** zhurong has joined #openstack-smaug | 13:21 | |
| *** zhurong has quit IRC | 13:43 | |
| *** zhurong has joined #openstack-smaug | 13:44 | |
| *** zhurong has quit IRC | 14:06 | |
| *** mingyu has joined #openstack-smaug | 14:06 | |
| *** zhurong has joined #openstack-smaug | 14:06 | |
| *** yamamoto has joined #openstack-smaug | 14:07 | |
| *** mingyu has quit IRC | 14:10 | |
| *** yamamoto has quit IRC | 14:13 | |
| openstackgerrit | Saggi Mizrahi proposed openstack/smaug: Support file system based bank implementation https://review.openstack.org/355956 | 14:15 |
| openstackgerrit | Saggi Mizrahi proposed openstack/smaug: Support file system based bank implementation https://review.openstack.org/355956 | 14:16 |
| xiangxinyong | saggi: thanks | 14:27 |
| openstackgerrit | xiangxinyong proposed openstack/smaug-dashboard: Fix Protection Plan Creation Failed https://review.openstack.org/355976 | 14:31 |
| *** yuval has quit IRC | 15:01 | |
| openstackgerrit | Saggi Mizrahi proposed openstack/smaug: Bank indexing interface https://review.openstack.org/356011 | 15:25 |
| openstackgerrit | Saggi Mizrahi proposed openstack/smaug: Bank indexing interface https://review.openstack.org/356011 | 15:27 |
| *** zhurong has quit IRC | 15:37 | |
| *** gouthamr has quit IRC | 20:33 | |
| *** yamamoto has joined #openstack-smaug | 21:08 | |
| *** yamamoto has quit IRC | 21:15 | |
| *** gouthamr has joined #openstack-smaug | 22:39 | |
| *** gouthamr_ has joined #openstack-smaug | 22:40 | |
| *** gouthamr has quit IRC | 22:44 | |
| *** gouthamr_ is now known as gouthamr | 22:44 | |
| *** yamamoto has joined #openstack-smaug | 22:46 | |
| *** zhurong has joined #openstack-smaug | 23:49 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!