Daniel Nephin
Daniel Nephin
Another option would be to change the KeyID to the primary key, and remove the `id` column. Or even rename the `keyID` column to `id`. That might make the migration...
Another option would be to change the KeyID to the primary key, and remove the `id` column. Or even rename the `keyID` column to `id`. That might make the migration...
> but keeping support for Kubernetes while in this mode seems unnecessary The goal here isn't to support different kinds of connectors, but to allow the kubernetes connector to run...
I think it's not specific to orgs. It happens with any call to `data.add` that is outside a transaction, which is the case for all data tests right now.
I think one nice way of solving this would be to remove the need for savepoints. I think if we use `ON CONFLICT DO NOTHING` (https://www.postgresql.org/docs/current/sql-insert.html#SQL-ON-CONFLICT) we can remove the...
Still relevant, and becomes even more important to replace with the introduction of orgs, as it makes caching more difficult and less effective.
This was done in 7a769470d8d349143098d779f3053d04396ca9a3. Thank you @ssoroka !
Still relevant
Fixed by #3299, pprof handler now has a separate read-only txn.
I've started some work on a set of architecture diagrams here: https://github.com/infrahq/infra/blob/dnephin/docs-architecture/docs/architecture.md