The psycopg2 package isn't available for PyPy. This commit adds a check
if the runtime is PyPy, and if it is uses psycopg2cffi module in favor
of psycopg2. This is almost a drop-in replacement, except for one place
where an additional cast to string is required.
We poked the notifier before updated the current token for the cache
invalidation stream. This mean that sometimes the update wouldn't be
sent until the next time a cache was invalidated.
For each request, track the amount of time spent waiting for a db
connection. This entails adding it to the LoggingContext and we may as well add
metrics for it while we are passing.
wrap the call to _simple_upsert_txn in a loop so that we retry on an
integrityerror: this means we can avoid locking the table provided there is an
unique index.
Add db_conn parameters to the `__init__` methods of the *Store classes, so that
they are all consistent, which makes the multiple inheritance work correctly
(and so that we can later extract mixins which can be used in the slavedstores)
This is due to the fact that we prefilled caches using txn.call_after,
which always gets called including on error.
We fix this by making txn.call_after only fire when a transaction
completes successfully, which is what we want most of the time anyway.
administrators can now:
- Set displayname of users
- Update user avatars
- Search for users by user_id
- Browse all users in a paginated API
- Reset user passwords
- Deactivate users
Helpers for doing paginated queries has also been added to storage
Signed-off-by: Morteza Araby <morteza.araby@ericsson.com>
Some tables, like device_inbox, take a long time to query at startup for
the stream change cache prefills. This is likely because they are slower
growing streams and so are more fragmented on disk. For now, lets pull
fewer entries out to make startup quicker.
In future, we should add a better index to make it even faster.
