| r1chardj0n3s | krotscheck: I assume that the username uniqueness change is related to ongoing work in the launchpad migration? | 00:01 |
|---|---|---|
| r1chardj0n3s | as in, there are changes in the migration code that aren't reflected in uniqueness change set | 00:01 |
| r1chardj0n3s | (I'm currently looking at LaunchpadWriter) | 00:02 |
| persia | If a +2 was added, is there any point to additional +1s? | 00:10 |
| r1chardj0n3s | persia: given I previously -1'd I figure it's only polite :) | 00:11 |
| openstackgerrit | Michael Krotscheck proposed a change to openstack-infra/storyboard: Fixed launchpad import bugs. https://review.openstack.org/133542 | 00:15 |
| openstackgerrit | Michael Krotscheck proposed a change to openstack-infra/storyboard: Permit name conflicts in users table. https://review.openstack.org/133571 | 00:15 |
| krotscheck | Ok, I think I have the pieces figured out now. | 00:15 |
| persia | r1chardj0n3s: In that situatiom. I'd agree. I was just wondering about procedure. | 00:16 |
| r1chardj0n3s | yup :) | 00:16 |
| krotscheck | r1chardj0n3s: Wait a sec, what’s up with LaunchpadWriter? | 00:16 |
| krotscheck | Oh, right | 00:17 |
| krotscheck | argh | 00:17 |
| r1chardj0n3s | krotscheck: write_user maps username to openid | 00:17 |
| krotscheck | hrm | 00:17 |
| r1chardj0n3s | yeah | 00:17 |
| krotscheck | Good catch | 00:17 |
| persia | I'm a bit surprised that "Incomplete (with response)" maps to 'invalid', as I usually only saw that in cases where the reporter didn't have an LP ACL grant, but it is a corner case. | 00:18 |
| krotscheck | r1chardj0n3s: You should -1 that | 00:23 |
| r1chardj0n3s | krotscheck: ok | 00:24 |
| * krotscheck tries to figure out a different way of uniquely identifying launchpad users :/ | 00:29 | |
| persia | All valid and current accounts should have a unique ~ entry in their OpenID URL. | 00:30 |
| persia | e.g. the "persia" in "https://launchpad.net/~persia" | 00:32 |
| persia | Although this doesn't map to person, just LP ID | 00:32 |
| persia | Err, LP *account* | 00:32 |
| persia | krotscheck: Rather then wracking your brain, perhaps try #launchpad. I'm happy to prod discussion, if that helps. | 00:37 |
| krotscheck | persia: Yeah, already found that and am running a test migration now to see how that works out. | 00:37 |
| * persia delays the attempt to invoke wgrant | 00:40 | |
| krotscheck | Wow, these ID’s are so inconsistent... | 00:42 |
| persia | How do you mean? | 00:42 |
| krotscheck | SOmetimes, the ~ name is a self-set username, sometimes it’s a firstname/lastname combo, somtimes it’s first initial last name, sometimes it’s first name + random characters. | 00:45 |
| persia | It is one of a) Debian username, b) self-selected value, c) something parsed from some commit message in some VCS mangled to not conflict with a & b | 00:46 |
| krotscheck | That seems sane. | 00:46 |
| * krotscheck might be sarcastic | 00:46 | |
| persia | Note that there's some historical messiness there, in that folk self-selecting after the VCS import thing happened had a restricted set, and newer Debian developers may not have gotten their name. | 00:46 |
| persia | All of it seemed like a good idea at the time :) | 00:47 |
| persia | (at least to some folk) | 00:47 |
| krotscheck | Either way, the point is that in the loader scripit (as r1chardj0n3s pointed out) the hash key to resolve the openid url was in no way related to the input parameter, so I fixed that. | 00:47 |
| openstackgerrit | Michael Krotscheck proposed a change to openstack-infra/storyboard: Permit name conflicts in users table. https://review.openstack.org/133571 | 00:47 |
| krotscheck | ‘cause really, if we’re going to say: We use this variable to get an openid and don’t want to repeat ourselves, storing it on a different variable makes NO sense. | 00:48 |
| r1chardj0n3s | so we want to try to retain that uinique user_link value in the import? | 00:49 |
| persia | It is the only thing LP promises to be unique | 00:50 |
| r1chardj0n3s | seems like it'd be a good idea to retain even for just-in-case | 00:50 |
| persia | Pre-OpenID, that was the token we used to track ability to do things in LP (so an app would check the permissions for a user based on LP group, using the LP ID as a unique string) | 00:51 |
| krotscheck | It really feels like we’ve got two unique identifiers here, the web_link and the openid. | 00:52 |
| krotscheck | Unfortunately launchpad has demonstrated that occasionally the openid can change. | 00:53 |
| persia | Or be reassigned. | 00:53 |
| krotscheck | RIght. | 00:53 |
| krotscheck | WHICH IS STUPID. | 00:53 |
| persia | It's probably safer to track web_link | 00:53 |
| persia | Note that this *can also* change, but the process is a bit more complex | 00:53 |
| krotscheck | That would require a rather aggressive change to our data model. | 00:53 |
| krotscheck | Ok, so what you’re saying is that launchpad doesn’t actually provide any kind of reliable unique idneifier for a user? | 00:54 |
| r1chardj0n3s | I think user_link is it | 00:54 |
| r1chardj0n3s | er web_link | 00:54 |
| persia | I'm saying that we used to use the LP ID (the string after the ~ in home pages) to identify, but that the LP UI provides a way to adjust that. | 00:54 |
| persia | If we want a real answer, I think we need to ask in #launchpad. I'm happy to ask, and know folk there, but need to sleep soon given the timezone in which I work tomorrow, so think it would be good to have a wider audience follow up my initial queries. | 00:55 |
| krotscheck | Right | 00:55 |
| krotscheck | I have to go as well, let’s restart this in the morning. | 00:56 |
| * krotscheck will ponder. | 00:56 | |
| krotscheck | and this discussion should probably end up in infra | 00:56 |
| krotscheck | Toodles everyone | 00:56 |
| r1chardj0n3s | cya krotscheck | 00:56 |
| persia | The folk with the answers are all in the antipodes, so it has to be tomorrow night | 00:56 |
| persia | (for folk on the east coast of the pacific ocean, at least) | 00:56 |
| persia | Or, I suppose, late afternoon ,depending on one's latitude | 00:57 |
| *** Viswanath has joined #storyboard | 02:23 | |
| *** miqui has joined #storyboard | 02:24 | |
| *** Viswanath has quit IRC | 02:26 | |
| *** timrc is now known as timrc-afk | 02:32 | |
| *** timrc-afk is now known as timrc | 03:27 | |
| *** timrc is now known as timrc-afk | 03:28 | |
| *** miqui has quit IRC | 05:11 | |
| *** k4n0 has joined #storyboard | 05:39 | |
| *** jtomasek has joined #storyboard | 07:24 | |
| *** alexismonville has joined #storyboard | 08:49 | |
| *** alexismonville has quit IRC | 09:24 | |
| *** yolanda has quit IRC | 09:46 | |
| *** yolanda has joined #storyboard | 10:05 | |
| *** miqui has joined #storyboard | 13:50 | |
| *** mattfarina has joined #storyboard | 14:49 | |
| *** openstackgerrit has quit IRC | 14:55 | |
| *** k4n0 has quit IRC | 15:27 | |
| *** timrc-afk is now known as timrc | 16:03 | |
| krotscheck | So, examining our use case of migration, I’m not concerned with storing the openid or the launchpad ~ name. | 16:09 |
| krotscheck | Migrations are likely to only happen at discrete points, and unless the ~ names or the openids chnage in that timeframe, we stop caring after the migration is complete. | 16:09 |
| krotscheck | Now, we may have some odd situations because we’re migrating in batches, but the worst that’ll happen there is that we’ll end up with some duplicate user records that willr equire a manual merge of those records. | 16:10 |
| krotscheck | And we’ll have to support something like that anyway. | 16:10 |
| krotscheck | ttx: If you’re around, we’ve got a couple of reviews for ya. The review chain starts here: https://review.openstack.org/#/c/133542/2 | 16:13 |
| persia | So you are giving up on actual uniqueness, figuring that the chance of change in a migration timeframe is small? | 16:15 |
| krotscheck | Yep. And there is no serious downside that we can’t mitigate with features we need anyway. | 16:19 |
| persia | Makes sense. Are you planning to pull from one of the less-used LP instances, or the normal one? | 16:20 |
| ttx | krotscheck: on it | 16:34 |
| krotscheck | ttx: Thanks | 16:34 |
| ttx | krotscheck: quick question: what purpose does line 29 serve in https://review.openstack.org/#/c/133571/5/storyboard/db/migration/alembic_migrations/versions/029_strip_user_constraints.py ? | 16:37 |
| krotscheck | ttx: None! | 16:37 |
| ttx | ah! | 16:37 |
| krotscheck | ttx: Just a sec. | 16:37 |
| * ttx holds | 16:38 | |
| krotscheck | ttx: Fixed | 16:41 |
| ttx | +2 | 16:42 |
| ttx | the first one merged | 16:42 |
| krotscheck | woot | 16:43 |
| * krotscheck waits for jenkins to go through before poking fungi one last time. | 16:44 | |
| *** Viswanath has joined #storyboard | 16:53 | |
| *** Viswanath has quit IRC | 16:56 | |
| *** mgagne has left #storyboard | 17:44 | |
| *** openstackgerrit has joined #storyboard | 17:48 | |
| *** yolanda has quit IRC | 19:14 | |
| *** yolanda has joined #storyboard | 19:32 | |
| *** yolanda has quit IRC | 19:32 | |
| *** yolanda has joined #storyboard | 19:32 | |
| *** yolanda_ has joined #storyboard | 19:32 | |
| openstackgerrit | A change was merged to openstack-infra/storyboard: Permit name conflicts in users table. https://review.openstack.org/133571 | 19:43 |
| *** yolanda_ has quit IRC | 20:58 | |
| *** yolanda has quit IRC | 20:59 | |
| krotscheck | Ok, so infra migration is in progress. | 21:00 |
| r1chardj0n3s | \o/ | 21:00 |
| *** Viswanath has joined #storyboard | 21:09 | |
| *** Viswanath has quit IRC | 21:12 | |
| *** miqui has quit IRC | 21:22 | |
| *** r1chardj0n3s is now known as r1chardj0n3s_afk | 22:17 | |
| *** r1chardj0n3s_afk is now known as r1chardj0n3s | 22:44 | |
| *** miqui has joined #storyboard | 22:50 | |
| *** mattfarina has quit IRC | 23:00 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!