| *** tosky has quit IRC | 00:04 | |
| *** lifeless_ has joined #storyboard | 00:46 | |
| *** lifeless has quit IRC | 00:46 | |
| *** ffmancerajv has joined #storyboard | 00:54 | |
| *** ffmancerajv has quit IRC | 00:54 | |
| EmilienM | diablo_rojo: when can we start migrating tripleo/ui bugs to sb? | 02:29 |
|---|---|---|
| *** jamesmcarthur has joined #storyboard | 03:13 | |
| *** jamesmcarthur has quit IRC | 04:18 | |
| *** jamesmcarthur has joined #storyboard | 07:18 | |
| *** jamesmcarthur has quit IRC | 07:23 | |
| *** jtomasek has joined #storyboard | 08:05 | |
| *** tosky has joined #storyboard | 09:21 | |
| *** jamesmcarthur has joined #storyboard | 13:47 | |
| *** jamesmcarthur has quit IRC | 13:47 | |
| *** jamesmcarthur has joined #storyboard | 13:48 | |
| ttx | FWIW there are two Thierry Carrez in StoryBoard, the one with ID 2394 should probably be dropped | 14:59 |
| ttx | I wonder where that one comes from | 14:59 |
| clarkb | I think there are two clarkb's too | 15:00 |
| clarkb | my hunch is maybe as part of the lp -> ubuntu one openid move? | 15:01 |
| tosky | oh, also two Luigi Toscano | 15:02 |
| persia | In general, storyboard probably needs some features to merge things. Duplicate users are annoying, but there are also duplicate stories. | 15:03 |
| tosky | I noticed after the import of sahara stories, but I'm not sure if it happened also before | 15:03 |
| diablo_rojo | ttx, clarkb yeah noticed that was happening last week. Its at the top of my list of things to look at. | 15:03 |
| persia | diablo_rojo: The import script matches users based on OpenID URL, right? Would that mean that as we import older stories, we end up with older URLs sometimes? | 15:04 |
| diablo_rojo | EmilienM, we can probably do that this week when we migrate Ironic on friday? | 15:04 |
| EmilienM | diablo_rojo: sure, just let us know | 15:05 |
| EmilienM | mwhahaha: ^ | 15:05 |
| EmilienM | mwhahaha: less bugs in Launchpad :P | 15:05 |
| diablo_rojo | EmilienM, mwhahaha I will chat with fungi about it today and see what time he has on Friday to do this. | 15:06 |
| fungi | i'm probably open on friday, but we're also scheduling project renaming for a couple projects during the infra meeting later today and may set that for friday | 15:07 |
| diablo_rojo | persia, yes it matches based on OpenIDs...I could see that potentially being the issue. | 15:07 |
| diablo_rojo | I haven't found a pattern yet so maybe thats it. | 15:08 |
| fungi | right, if lp has duplicate accounts with different openid urls that could explain it | 15:08 |
| fungi | i'll take a look at some of the examples mentioned above and see if i can spot any interesting clues | 15:08 |
| persia | My understanding is that new users in StoryBoard are created when a) someone logs in, and b) a migration runs and a user is credited with comments. | 15:09 |
| persia | I don't know of any particular reason that the OpenID URLs for (a) and (b) would necessarily be a perfect match. | 15:09 |
| persia | (and we'll probably have to update everything again if people want to be able to use a different OpenID provider, given the current model) | 15:10 |
| SotK | interesting, there is only one Adam Coldrick | 15:41 |
| SotK | I guess this adds more weight to the theory they were added by the migration script | 15:42 |
| tosky | I have to add that at some point launchpad ID had two accounts with my name; I could see that from gerrit, as people could not easily add me as reviewer | 15:48 |
| tosky | then the duplicate was removed, so maybe it's connected to that, even if it was long time ago | 15:49 |
| persia | The trick with a collapse-duplicate-user utility is that we know there are some folk with the same name in StoryBoard already. Some of us have unique-to-openstack names, but some do not. | 16:01 |
| fungi | yeah, gerrit has a similar new-openid-means-new-account model, and we have no end of challenge with duplicate accounts there from changes in ubuntu sso triggering new openids | 16:38 |
| persia | I know the Storyboard account model was a temporary stopgap chosen for ease of implementation to start, which has never been revisited. I wonder if that also applies in other cases. | 16:41 |
| persia | A sensible model is to have 1->many relation between identity and OpenID URL, where the user is expected to log in with at least one registered OpenID URL to add more, but that requires complex UI, instructions, etc. | 16:41 |
| *** jamesmcarthur has quit IRC | 16:47 | |
| *** jamesmcarthur has joined #storyboard | 18:12 | |
| *** diablo_rojo has quit IRC | 18:22 | |
| fungi | yeah, and usually the way people find out they have another openid is to log in with the wrong one and end up with two accounts | 18:30 |
| fungi | so ultimately you need to be able to merge those afterward | 18:30 |
| fungi | "i can't remember which way i logged in last time, let's try this one" | 18:30 |
| *** diablo_rojo has joined #storyboard | 18:37 | |
| tosky | ... that said, can anyone merge my two accounts on storyboard :) or are you looking for a more general solutions? | 18:55 |
| *** jamesmcarthur has quit IRC | 19:11 | |
| *** jamesmcarthur has joined #storyboard | 19:13 | |
| *** diablo_rojo has quit IRC | 19:44 | |
| persia | tosky: We're discussing a more general solution. Unless I misunderstand, there are no tools to merge accounts (and the necessary API bits aren't exposed), so someone would have to write a script to manually change all the values in the database. Given the effort involved, it is probably easier to write a semi-manual script, and not much harder to add an admin API, at which point SB admins would be able to merge the accounts. | 19:48 |
| *** jdandrea_ has quit IRC | 19:56 | |
| *** tellesnobrega has quit IRC | 20:11 | |
| *** jamesmcarthur has quit IRC | 20:55 | |
| *** jamesmcarthur has joined #storyboard | 21:03 | |
| *** jamesmcarthur has quit IRC | 21:04 | |
| *** jamesmcarthur has joined #storyboard | 21:04 | |
| *** diablo_rojo has joined #storyboard | 21:21 | |
| *** jamesmcarthur has quit IRC | 21:39 | |
| *** ssbarnea has quit IRC | 21:46 | |
| *** ssbarnea has joined #storyboard | 21:52 | |
| *** diablo_rojo has quit IRC | 22:13 | |
| *** diablo_rojo has joined #storyboard | 22:14 | |
| fungi | yeah, i was going to do some database queries to see if i can guess at a general cause (or perhaps more than one) | 22:16 |
| *** jamesmcarthur has joined #storyboard | 22:18 | |
| fungi | so looking at ttx's example, his id#6 uses a https://login.ubuntu.com/ openid (which we converted from login.launchpad.net) while his id#2394 uses a https://login.launchpad.net/ id created 2018-03-16 after we converted the old accounts | 22:18 |
| fungi | i wonder if the import process is repopulating our users table with old launchpad.net openids | 22:19 |
| fungi | we could do a simple url translation on import, but still need to do user id substitution in all the tables where that gets reused | 22:20 |
| fungi | tosky: same deal with your original id#859 using https://login.ubuntu.com/ and your duplicate id#2407 using https://login.launchpad.net/ | 22:21 |
| fungi | 859 created on 2015-02-26, 2407 created on 2018-03-16 | 22:22 |
| fungi | looks like a pattern | 22:23 |
| tosky | and 2018-03-16 is the Sahara import date | 22:23 |
| fungi | luckily, the path portion of the openid urls is identical so we can deduplicate based on that just need to work out all the tables which need user id updates | 22:23 |
| fungi | i can almost certainly brute-force a solution in the database, but i'll give it a day or two for sb veterans to weigh in with suggestions first | 22:24 |
| fungi | worst case i'll do some hacky etl in python | 22:25 |
| diablo_rojo | fungi, I think if we can update the migration script that would be better, but I am fine with a brute force thing to clean up the state of things now | 22:25 |
| fungi | diablo_rojo: well, it's both | 22:25 |
| fungi | we need to correct the data which is already in the database, and adjust the migration script to prevent any recurrence | 22:26 |
| fungi | fixing in the migration script is probably very easy, just need to s/login.launchpad.net/login.ubuntu.com/ on the openid field | 22:26 |
| diablo_rojo | So changes somewhere around here I would guess? https://github.com/openstack-infra/storyboard/blob/master/storyboard/migrate/launchpad/writer.py#L140 | 22:27 |
| fungi | deduplicating accounts in the db is the tricky bit | 22:27 |
| fungi | for the already duplicated accounts | 22:27 |
| fungi | yeah, somewhere in that vicinity before the row in the users table is committed should work | 22:28 |
| diablo_rojo | so replace login.ubuntu.com for where login.launchpad.net happens? | 22:28 |
| fungi | yeah, it looks like the openid being given back by the lp api has the former and we want to perform string substitution on the result to transform it to the latter | 22:30 |
| fungi | er, other way around from what i said | 22:30 |
| fungi | replace login.launchpad.net in what the lp api gives back with login.ubuntu.com before committing to the db | 22:30 |
| diablo_rojo | Will push the patch in a minute. Making the changes now. | 22:32 |
| fungi | diablo_rojo: oh, unrelated tidbit i promised i would pass along... during the zuul meeting yesterday where we discussed remaining hurdles for using sb for embargoed vulnerability reports it was pointed out that there's an issue getting e-mail notification for updates to private stories: http://eavesdrop.openstack.org/meetings/zuul/2018/zuul.2018-03-19-22.03.log.html#l-32 | 22:32 |
| diablo_rojo | fungi, we also should probably get this merged https://review.openstack.org/#/c/532706/ | 22:33 |
| fungi | corvus has been working on getting sufficient regression testing in place so that we can test the fix for it so that it doesn't stop working later, but is running into some roadblocks | 22:33 |
| diablo_rojo | Since its been out there for a while and this new migration change will depend on that one. | 22:34 |
| fungi | reviewing now | 22:34 |
| diablo_rojo | fungi, thanks | 22:34 |
| diablo_rojo | fungi, wrt the vmt processes and emails- I saw corvus's WIP patch go up, but havent had time to look at it in detail yet. | 22:35 |
| diablo_rojo | I can add a ask onto the private stories story about the email notification stuff- or we can split it out as its own story. | 22:36 |
| fungi | diablo_rojo: did you want to address persia's remaining comment on 532706? i'm good either way though it does seem like a (minor) improvement | 22:36 |
| fungi | it's already less judgemental than "anonymous coward" (which is what gerrit displays) so i'm not to concerned either way | 22:37 |
| *** openstackgerrit has joined #storyboard | 22:37 | |
| openstackgerrit | Kendall Nelson proposed openstack-infra/storyboard master: Migration Error with Suspended User https://review.openstack.org/532706 | 22:37 |
| diablo_rojo | fungi, edit published :) | 22:37 |
| corvus | i've been working on that for about the past hour... i still don't understand the database system at all :( | 22:38 |
| diablo_rojo | I am totally behind calling them something else :) | 22:38 |
| corvus | i think my supposition was wrong -- it's not storing the data in the db where the problem occurs.. it looks like it's actually the query to use it. which raises the question of how the permissions query works in production. that makes me think perhaps it's related to the session handling (which is different for the api and in the test suite). | 22:40 |
| *** jamesmcarthur has quit IRC | 22:41 | |
| openstackgerrit | Kendall Nelson proposed openstack-infra/storyboard master: Duplicating Users at Migration Time https://review.openstack.org/554726 | 22:51 |
| diablo_rojo | I didn't make it dependent on the other migration script edit, but the merge conflict will be super simple to clean up. | 22:52 |
| diablo_rojo | There you go fungi ^^ | 22:52 |
| fungi | thanks, i'll review both | 22:53 |
| diablo_rojo | fungi, thanks :) | 22:54 |
| persia | For user deduplication, we probably want to consider both the cases of users already duplicated and users who are not currently duplicated but for whom we have recorded the incorrect OpenID URL in the database, such that they will end up becoming duplicate the first time they authenticate to StoryBoard. | 22:55 |
| persia | This is related to the user duplication cleanup in the migration script, but more about whatever needs to happen in the DB first. | 22:55 |
| fungi | right, i would consider both of those "duplicated" for purposes of data cleanup | 22:55 |
| persia | As would I: the key is that in some case there is only one "duplicate" record (where the correct record does not exist). | 22:56 |
| fungi | i'll make sure we don't have any remaining login.launchpad.net openids at the end of whatever cleanup is performed | 22:56 |
| persia | Excellent. Just wanted to make sure that detail was caught when the migration happens, so we wouldn't have to do it again. | 22:57 |
| fungi | but thanks for the reminder, i might have forgotten to design the cleanup to account for those | 22:57 |
| openstackgerrit | Jeremy Stanley proposed openstack-infra/storyboard master: Duplicating Users at Migration Time https://review.openstack.org/554726 | 23:01 |
| fungi | diablo_rojo: ^ lgtm but heavy edit of the commit message to provide clearer explanation/history on the issue | 23:01 |
| diablo_rojo | fungi, works for me :) | 23:02 |
| corvus | i'm hitting the wall again on the private story thing. i'm happy to continue working on it, but i need help from someone who understands the storyboard database api to continue. | 23:02 |
| diablo_rojo | Aside from lines not bring wrapped lol. | 23:02 |
| fungi | diablo_rojo: oh, ouch. i forget that gertty doesn't auto-wrap commit messages. i'll fix it | 23:02 |
| openstackgerrit | Jeremy Stanley proposed openstack-infra/storyboard master: Duplicating Users at Migration Time https://review.openstack.org/554726 | 23:03 |
| diablo_rojo | corvus, I am probably only a little more knowledgeable than you. SotK is your man for helping with that. | 23:05 |
| diablo_rojo | I am happy to be a rubber duck though. | 23:05 |
| diablo_rojo | fungi, looks good now :) | 23:05 |
| corvus | diablo_rojo, SotK: i'll try to pop in here first thing tomorrow morning then | 23:05 |
| fungi | diablo_rojo: if that one's not merged come friday, remind me to single-core approve it (and make sure i've already fixed up the db) before i start the import for tripleo | 23:06 |
| fungi | better i don't make the mess even bigger | 23:07 |
| * diablo_rojo makes note to remind fungi | 23:07 | |
| fungi | though honestly, it'll be an etl script querying and updating the db no matter how many dupes i need to contend with | 23:07 |
| fungi | so not a huge deal either way | 23:07 |
| diablo_rojo | fungi, what time are you planning on running the ironic and tripleO UI teams migrations? | 23:10 |
| fungi | i have no idea | 23:11 |
| fungi | probably in the latter utc half of the day if that works for everyone | 23:11 |
| fungi | since those are my typical waking hours | 23:11 |
| diablo_rojo | Works for me. I just wanted to know if I had to get up early to be here for you/remind you of things. | 23:16 |
| openstackgerrit | Merged openstack-infra/storyboard master: Migration Error with Suspended User https://review.openstack.org/532706 | 23:16 |
| *** jamesmcarthur has joined #storyboard | 23:24 | |
| *** jamesmcarthur has quit IRC | 23:29 | |
| *** tosky has quit IRC | 23:34 | |
| fungi | nope! i'm a lazy^H^Hte riser | 23:46 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!