Closed andymeneely closed 10 years ago
Assigning this to @bspates
@andymeneely, which is the current production db? Is it chromium_real? I just wanted to get a good sample of emails to test with.
Sadly, I made a mistake and blew away production last night. It'll rebuild tonight and you should have access to it.
Looks like I don't have permission to query the chromium_real or chromium_real2 dbs
Looks like there is a google group under committers@chromium.org, that lists all the commiters to the project. Aparently having write access is based of them scraping this list for the email. https://code.google.com/p/chromium/issues/detail?id=237072, this conversation talks about eventually allowing people with non "@chromium.org" addresses to commit. This seems confusing considering the data we've gathered.
@andymeneely forgot to put your name in the last two comments, but I need read access for the prod data, and checkout the comment before this one cause I'm a little confused by some stuff i found.
Oops, right. Just work with chromium_real
and not chromium_real2
, you have read permissions but not write.
Here are the measures we discussed that we will take:
gtempaccount
email, then we need to modify search_and_add to simply convert the email address to the original email. For example, dgozman%chromium.org@gtempaccount.com
should be converted to dgozman@chromium.org
search_and_add
return nil, and have all callers do a nil check.In a separate consolidation routine AFTER all data has been loaded,
brettw@chromium.org
Once we do all of the above, let's start a manual investigation of the ones we have left. We can do a manual mapping of our own.
@andymeneely looks like i lost permission to the chromium_real db again .
Shoot. Looks like this is going to happen every day because of the way I do the rename from chromium_real2
to chromium_real
. I just gave you admin. Be careful.
From the prod logs it looks like some gtemp accounts snuck through somehow, also it seems to be rejecting some perfectly good email address, will investigate.
Closed for now since we've gotten it all tested and running with the new performance optimizations. Other issue #143 open.
Take a look at our development data code review 5831706594508800 - looks like we have two reviewers that look like the same person:
skaslev@chromium.org skaslev@google.com
How often does this happen? What types of disamiguation should we do? Maybe something as a part of rake run:consolidate?