| *** daemontool has joined #openstack-freezer | 02:25 | |
| *** c00281451 is now known as chenzeng | 03:32 | |
| *** dschroeder has quit IRC | 03:56 | |
| *** Slashme has joined #openstack-freezer | 06:25 | |
| *** Slashme has quit IRC | 06:31 | |
| *** samuelBartel has joined #openstack-freezer | 08:29 | |
| *** EinstCrazy has quit IRC | 09:01 | |
| *** EinstCrazy has joined #openstack-freezer | 09:02 | |
| *** EinstCrazy has quit IRC | 09:07 | |
| *** openstackgerrit has quit IRC | 09:17 | |
| *** openstackgerrit has joined #openstack-freezer | 09:18 | |
| *** pbourke has quit IRC | 10:00 | |
| *** pbourke has joined #openstack-freezer | 10:06 | |
| *** EinstCrazy has joined #openstack-freezer | 10:22 | |
| *** EinstCrazy has quit IRC | 10:35 | |
| *** EinstCrazy has joined #openstack-freezer | 10:36 | |
| *** EinstCrazy has quit IRC | 10:40 | |
| openstackgerrit | Pierre Mathieu proposed openstack/freezer: Fixing database config file parsing https://review.openstack.org/276333 | 10:47 |
|---|---|---|
| openstackgerrit | Pierre Mathieu proposed openstack/freezer: Fixing database config file parsing https://review.openstack.org/276333 | 10:50 |
| *** jokke_ has quit IRC | 11:12 | |
| *** jokke_ has joined #openstack-freezer | 11:13 | |
| openstackgerrit | Eldar Nugaev proposed openstack/freezer: Switch cinder api to v2 and add --force and --incremental https://review.openstack.org/276685 | 11:18 |
| *** reldan_ has joined #openstack-freezer | 11:38 | |
| *** memogarcia has quit IRC | 11:41 | |
| *** reldan_ has quit IRC | 11:44 | |
| *** memogarcia has joined #openstack-freezer | 12:06 | |
| *** reldan_ has joined #openstack-freezer | 12:09 | |
| *** EinstCrazy has joined #openstack-freezer | 12:29 | |
| *** EinstCrazy has quit IRC | 12:36 | |
| *** EmilDi_ has joined #openstack-freezer | 12:38 | |
| *** daemontool has quit IRC | 12:38 | |
| *** reldan_ has quit IRC | 12:46 | |
| *** reldan_ has joined #openstack-freezer | 12:51 | |
| *** daemontool has joined #openstack-freezer | 13:11 | |
| *** Slashme has joined #openstack-freezer | 13:12 | |
| daemontool | Hi | 13:17 |
| memogarcia | hello | 13:21 |
| *** Slashme has quit IRC | 13:24 | |
| *** EmilDi_ has quit IRC | 13:31 | |
| *** EmilDi_ has joined #openstack-freezer | 13:33 | |
| *** chenzeng has quit IRC | 13:34 | |
| daemontool | EmilDi, ping | 13:49 |
| daemontool | EmilDi_, ^^ | 13:50 |
| EmilDi_ | hi daemontool | 13:51 |
| *** Slashme has joined #openstack-freezer | 14:53 | |
| openstackgerrit | Fausto Marzi proposed openstack/freezer-api: Add Sphinx doc build https://review.openstack.org/276768 | 14:57 |
| *** Slashme has quit IRC | 15:02 | |
| openstackgerrit | Fausto Marzi proposed openstack/freezer-web-ui: Add Sphinx doc build https://review.openstack.org/276777 | 15:19 |
| daemontool | reldan_, as soon I'll finish couple of things for mitaka-3 | 15:27 |
| daemontool | we'll see the metadata | 15:27 |
| reldan_ | daemontool: Great. For v2 and incremental everything is ok. But we need to add additional parameters to cli https://review.openstack.org/#/c/276685/1 | 15:28 |
| daemontool | reldan_, looking | 15:43 |
| openstackgerrit | Fausto Marzi proposed openstack/freezer: Add Sphinx doc build https://review.openstack.org/276488 | 15:53 |
| daemontool | reldan_, if you take a look at http://docs.openstack.org/admin-guide-cloud/blockstorage_volume_backups.html | 16:00 |
| daemontool | in some point it says "Because volume backups are dependent on the Block Storage database, you must also back up your Block Storage database regularly to ensure data recovery." | 16:00 |
| daemontool | in the following note | 16:00 |
| daemontool | it explains how to do that by backing up also the metadata | 16:01 |
| daemontool | I think we should probably include that | 16:01 |
| reldan_ | I can store metadata, but what I should do during restore then? | 16:02 |
| daemontool | I think we have to import that following docs.openstack.org/admin-guide-cloud/blockstorage_volume_backups_export_import.html#volume-backups-export-import | 16:03 |
| reldan_ | From my point of view - it is something a little bit different. It is usefull during the complete loosing of object-storage database. | 16:05 |
| reldan_ | I can store metadata somewhere | 16:05 |
| reldan_ | It is not problem - but it looks like a hack | 16:05 |
| reldan_ | because I don’t manage cindernative - I should create a folder in additional storage (where?) and create mechanism for restoring | 16:06 |
| reldan_ | only for cindernative | 16:06 |
| reldan_ | only for purposes of fatal failure of openstack | 16:06 |
| *** m3m0 has quit IRC | 16:07 | |
| openstackgerrit | Merged openstack/freezer: Add check for name=='__main__' https://review.openstack.org/266552 | 16:08 |
| *** dschroeder has joined #openstack-freezer | 16:10 | |
| openstackgerrit | Merged openstack/freezer: Fix configuration option restore_from_host https://review.openstack.org/275361 | 16:12 |
| openstackgerrit | Merged openstack/freezer: Fix a mistake in 'plugin.sh' https://review.openstack.org/270315 | 16:12 |
| daemontool | reldan_, failure of cinder db I think is a good enough purpose | 16:12 |
| daemontool | also if we want to restore on a different openstack instance | 16:12 |
| reldan_ | daemontool: Good. But where I should store this metadata? | 16:13 |
| reldan_ | for cindernative I have no storage | 16:13 |
| daemontool | in one of the media storage we support | 16:13 |
| daemontool | there's no other option I think | 16:13 |
| daemontool | or | 16:13 |
| daemontool | in the freezer api, but in that case we'll not be able to restore in case of cinder db failure | 16:13 |
| daemontool | if the freezer api are also not available | 16:14 |
| reldan_ | Let’s say I have this command freezerc --cindernative-vol-id c1580f0d-5414-48e4-9bf8-d8a5d40e166a | 16:14 |
| daemontool | I'd use the supported media storage | 16:14 |
| reldan_ | And I don’t specify any storage | 16:14 |
| daemontool | if no storage is specified it would be the default one, which is swift | 16:14 |
| daemontool | I think it make sense, so we offer some added value | 16:15 |
| daemontool | to the pure cinder backup | 16:15 |
| reldan_ | What if I have no swift, and cinder backend is ceph | 16:15 |
| reldan_ | but I have cinder | 16:15 |
| reldan_ | in this case we are unable to perform backup, aren’t we? | 16:16 |
| daemontool | so the case would be | 16:16 |
| daemontool | we do not have swift | 16:16 |
| daemontool | ssh node | 16:16 |
| reldan_ | Yes, only cindernative backup | 16:16 |
| daemontool | then we have local fs | 16:16 |
| daemontool | I see your point is good | 16:17 |
| daemontool | but I think that is an issue we should cover | 16:17 |
| daemontool | I know it complicates things a bit lol | 16:17 |
| reldan_ | Yes, because it is easy to implement, but hard to support | 16:17 |
| reldan_ | I can check if we have swift | 16:17 |
| daemontool | do you mean operationally? | 16:17 |
| daemontool | well | 16:17 |
| daemontool | in most of the cases | 16:18 |
| daemontool | in production | 16:18 |
| daemontool | you have swift | 16:18 |
| reldan_ | create a container “cindernative_backups/uuid” | 16:18 |
| reldan_ | and store there metadata | 16:18 |
| daemontool | yes | 16:18 |
| daemontool | that's a good approach | 16:18 |
| daemontool | we think a bit more on it | 16:18 |
| daemontool | but we can differenciate and provide added value if we do that | 16:18 |
| reldan_ | if I have no swift - I can store it in /tmp/cindernative_backups/uuid | 16:18 |
| daemontool | well | 16:18 |
| daemontool | $USER/.freezer/cindernative_backups/uuid | 16:19 |
| daemontool | I'll check, 1) swift, 2) ssh node 3) local fs | 16:19 |
| daemontool | and if none of the 3 are available | 16:19 |
| daemontool | we do the backup | 16:19 |
| daemontool | but log the limitations | 16:19 |
| daemontool | do you think is reasonable? | 16:20 |
| reldan_ | I think it is doable, but I doubt that it is a freezer problem %) | 16:22 |
| *** EmilDi_ has quit IRC | 16:22 | |
| daemontool | well, always wants to provide a way to restore | 16:22 |
| daemontool | if one or more openstack service is not available | 16:22 |
| daemontool | it's not our problem | 16:23 |
| daemontool | that's right | 16:23 |
| daemontool | but providing a solution is good after all | 16:23 |
| reldan_ | Yes, but it leads us from unified solution to the set of different mechanism (not related to each other) for every media | 16:24 |
| reldan_ | If I should consider possibility that OS just unable to keep metainformation, why I should be sure that I cannot lose swift | 16:25 |
| reldan_ | From my point of view freezer shouldn’t know much about OS and internal implementation. | 16:26 |
| reldan_ | And now I know that OS can lose it is own database | 16:26 |
| reldan_ | And prevent it in my code | 16:26 |
| daemontool | yes, that's why I think we should use as much as possible the existing os services api | 16:27 |
| daemontool | I totally agree | 16:27 |
| reldan_ | If cinder guys know that they can lose it’s own database. Probably they should store metadata in backup config file | 16:28 |
| reldan_ | Let’s say it the same, if freezer asks - please store all metadata about backup on you disk, because we can lose our database | 16:29 |
| daemontool | well I agree | 16:30 |
| daemontool | that's their limitation | 16:30 |
| reldan_ | But yes, sure I can implement this | 16:32 |
| daemontool | that will be the same I think for vm backups | 16:34 |
| daemontool | reldan_, do we want to add that to the bp? | 16:40 |
| daemontool | check also if zhangjn and EinstCrazy can be involved | 16:41 |
| reldan_ | daemontool: But for vm we don’t have native approach | 16:46 |
| *** samuelBartel has quit IRC | 16:46 | |
| reldan_ | or you mean snapshoots? | 16:46 |
| daemontool | we have backup api endpoint in nova, that uses snapshots internally | 16:48 |
| daemontool | problem is there's no incremental there | 16:48 |
| daemontool | cause cinder also uses snapshots, the incrementals are computed on the snapshots | 16:49 |
| reldan_ | I suppose snapshot is incremental already. If we don’t want to download it - we can have increments | 16:51 |
| daemontool | where in Nova? | 16:52 |
| daemontool | In qemu yes, I don't know if that feature is exported | 16:53 |
| daemontool | have to check | 16:53 |
| *** reldan_ has quit IRC | 17:15 | |
| *** reldan has joined #openstack-freezer | 17:40 | |
| *** pennerc has joined #openstack-freezer | 17:56 | |
| *** reldan has quit IRC | 17:56 | |
| *** reldan has joined #openstack-freezer | 18:02 | |
| *** reldan has quit IRC | 18:03 | |
| *** pbourke has quit IRC | 18:06 | |
| *** reldan has joined #openstack-freezer | 18:59 | |
| *** daemontool has quit IRC | 19:21 | |
| *** daemontool has joined #openstack-freezer | 19:25 | |
| *** memogarcia has quit IRC | 19:27 | |
| openstackgerrit | Fausto Marzi proposed openstack/freezer: Add Sphinx doc build https://review.openstack.org/276488 | 20:00 |
| openstackgerrit | Fausto Marzi proposed openstack/freezer-api: Add Sphinx doc build https://review.openstack.org/276768 | 20:09 |
| openstackgerrit | Fausto Marzi proposed openstack/freezer-web-ui: Add Sphinx doc build https://review.openstack.org/276777 | 20:13 |
| *** memogarcia has joined #openstack-freezer | 20:22 | |
| daemontool | Hi memogarcia | 20:23 |
| openstackgerrit | Fausto Marzi proposed openstack/freezer-web-ui: Add Sphinx doc build https://review.openstack.org/276777 | 20:23 |
| *** reldan has quit IRC | 20:53 | |
| *** pennerc has quit IRC | 21:44 | |
| *** reldan has joined #openstack-freezer | 22:17 | |
| *** daemontool has quit IRC | 22:30 | |
| *** reldan has quit IRC | 23:04 | |
| *** memogarcia has quit IRC | 23:15 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!