Superset access required for CommTech's work with Realtime Preview, can be revoked after that project is complete (~3 months from now).
| Subject | Repo | Branch | Lines +/- | |
|---|---|---|---|---|
| admin: Add samwilson to analytic privatedata group | operations/puppet | production | +2 -1 |
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | Samwilson | T290521 Real-time Preview: Data Questions | |||
| Resolved | • MMandere | T296161 Grant Access to PII in Superset for samwilson |
Hi @Samwilson: Can you please clarify as per https://wikitech.wikimedia.org/wiki/Analytics/Data_access#Dashboards_in_Superset_/_Hive_interfaces_(like_Hue)_that_do_access_private_data if you require access to dashboards with private data?
Noting https://ldap.toolforge.org/user/samwilson is already in the wmf group, so @Samwilson should be able to get past the login screen at superset.wikimedia.org and view some dashboards (that do not access private data).
This must be that kind of request as well then: https://wikitech.wikimedia.org/wiki/Analytics/Data_access#Dashboards_in_Superset_/_Hive_interfaces_(like_Hue)_that_do_access_private_data
Change 740826 had a related patch set uploaded (by MMandere; author: MMandere):
[operations/puppet@production] admin: Add samwilson to analytic privatedata group
Change 740826 merged by MMandere:
[operations/puppet@production] admin: Add samwilson to analytic privatedata group
@Samwilson you now should be able to access the private data. Please let us know if you face any challenges.
Thanks!
I'm still getting an error when I try to view FROM event.visualeditorfeatureuse:
Permission denied: user=samwilson, access=EXECUTE, inode="/wmf/data/event":analytics:analytics-privatedata-users …
I've tried logging out and in again but still get it. Does the permission change take some time to sync?
@Samwilson Great, you're welcome! Is there something else you did for it to start working?
No, I don't think so. I did try logging out and in again, but the fix came some time after that.
@Samwilson : It seems related to Puppet (our configuration management system) run times. Your update that it was still failing happened 21 minutes after Marc had merged the patch and our servers appliy configuration changes twice per hour (but distributed across the full 30 minute range to spread the load). You were simply unlucky :-)