Commit Graph

53 Commits

Author SHA1 Message Date
Erik Johnston
fa8934b175 Reduce serialization errors in MultiWriterIdGen (#8456)
We call `_update_stream_positions_table_txn` a lot, which is an UPSERT
that can conflict in `REPEATABLE READ` isolation level. Instead of doing
a transaction consisting of a single query we may as well run it outside
of a transaction.
2020-10-07 17:08:58 +01:00
Erik Johnston
695240d34a
Fix DB query on startup for negative streams. (#8447)
For negative streams we have to negate the internal stream ID before
querying the DB.

The effect of this bug was to query far too many rows, slowing start up
time, but we would correctly filter the results afterwards so there was
no ill effect.
2020-10-02 12:22:19 +01:00
Erik Johnston
b1433bf231
Don't table scan events on worker startup (#8419)
* Fix table scan of events on worker startup.

This happened because we assumed "new" writers had an initial stream
position of 0, so the replication code tried to fetch all events written
by the instance between 0 and the current position.

Instead, set the initial position of new writers to the current
persisted up to position, on the assumption that new writers won't have
written anything before that point.

* Consider old writers coming back as "new".

Otherwise we'd try and fetch entries between the old stale token and the
current position, even though it won't have written any rows.

Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>

Co-authored-by: Andrew Morgan <1342360+anoadragon453@users.noreply.github.com>
2020-09-29 16:42:19 +01:00
Erik Johnston
bd380d942f
Add checks for postgres sequence consistency (#8402) 2020-09-28 18:00:30 +01:00
Erik Johnston
3e87d79e1c
Fix schema delta for servers that have not backfilled (#8396)
Fixes #8395.
2020-09-25 09:58:32 +01:00
Erik Johnston
f112cfe5bb
Fix MultiWriteIdGenerator's handling of restarts. (#8374)
On startup `MultiWriteIdGenerator` fetches the maximum stream ID for
each instance from the table and uses that as its initial "current
position" for each writer. This is problematic as a) it involves either
a scan of events table or an index (neither of which is ideal), and b)
if rows are being persisted out of order elsewhere while the process
restarts then using the maximum stream ID is not correct. This could
theoretically lead to race conditions where e.g. events that are
persisted out of order are not sent down sync streams.

We fix this by creating a new table that tracks the current positions of
each writer to the stream, and update it each time we finish persisting
a new entry. This is a relatively small overhead when persisting events.
However for the cache invalidation stream this is a much bigger relative
overhead, so instead we note that for invalidation we don't actually
care about reliability over restarts (as there's no caches to
invalidate) and simply don't bother reading and writing to the new table
in that particular case.
2020-09-24 16:53:51 +01:00
Erik Johnston
cbabb312e0
Use async with for ID gens (#8383)
This will allow us to hit the DB after we've finished using the generated stream ID.
2020-09-23 16:11:18 +01:00
Erik Johnston
04cc249b43
Add experimental support for sharding event persister. Again. (#8294)
This is *not* ready for production yet. Caveats:

1. We should write some tests...
2. The stream token that we use for events can get stalled at the minimum position of all writers. This means that new events may not be processed and e.g. sent down sync streams if a writer isn't writing or is slow.
2020-09-14 10:16:41 +01:00
Erik Johnston
deedb91732
Fix MultiWriterIdGenerator.current_position. (#8257)
It did not correctly handle IDs finishing being persisted out of
order, resulting in the `current_position` lagging until new IDs are
persisted.
2020-09-08 14:26:54 +01:00
Richard van der Hoff
0dae7d80bf
Add more logging to debug slow startup (#8264)
I'm hoping this will provide some pointers for debugging
https://github.com/matrix-org/synapse/issues/7968.
2020-09-07 13:36:02 +01:00
Patrick Cloke
c619253db8
Stop sub-classing object (#8249) 2020-09-04 06:54:56 -04:00
Brendan Abolivier
9f8abdcc38
Revert "Add experimental support for sharding event persister. (#8170)" (#8242)
* Revert "Add experimental support for sharding event persister. (#8170)"

This reverts commit 82c1ee1c22.

* Changelog
2020-09-04 10:19:42 +01:00
Erik Johnston
82c1ee1c22
Add experimental support for sharding event persister. (#8170)
This is *not* ready for production yet. Caveats:

1. We should write some tests...
2. The stream token that we use for events can get stalled at the minimum position of all writers. This means that new events may not be processed and e.g. sent down sync streams if a writer isn't writing or is slow.
2020-09-02 15:48:37 +01:00
Erik Johnston
bbb3c8641c
Make MultiWriterIDGenerator work for streams that use negative stream IDs (#8203)
This is so that we can use it for the backfill events stream.
2020-09-01 13:36:25 +01:00
Erik Johnston
5649b7f3d0
Fix missing _add_persisted_position (#8179)
This was forgotten in #8164.
2020-08-27 13:20:34 +01:00
Erik Johnston
eba98fb024
Add functions to MultiWriterIdGen used by events stream (#8164) 2020-08-25 17:32:30 +01:00
Erik Johnston
2231dffee6
Make StreamIdGen get_next and get_next_mult async (#8161)
This is mainly so that `StreamIdGenerator` and `MultiWriterIdGenerator`
will have the same interface, allowing them to be used interchangeably.
2020-08-25 15:10:08 +01:00
Erik Johnston
c9c544cda5
Remove ChainedIdGenerator. (#8123)
It's just a thin wrapper around two ID gens to make `get_current_token`
and `get_next` return tuples. This can easily be replaced by calling the
appropriate methods on the underlying ID gens directly.
2020-08-19 13:41:51 +01:00
Erik Johnston
76d21d14a0
Separate get_current_token into two. (#8113)
The function is used for two purposes: 1) for subscribers of streams to
get a token they can use to get further updates with, and 2) for
replication to track position of the writers of the stream.

For streams with a single writer the two scenarios produce the same
result, however the situation becomes complicated for streams with
multiple writers. The current `MultiWriterIdGenerator` does not
correctly handle the first case (which is not an issue as its only used
for the `caches` stream which nothing subscribes to outside of
replication).
2020-08-19 10:39:31 +01:00
Erik Johnston
a7bdf98d01
Rename database classes to make some sense (#8033) 2020-08-05 21:38:57 +01:00
Richard van der Hoff
42509b8fb6 Use PostgresSequenceGenerator from MultiWriterIdGenerator
partly just to show it works, but alwo to remove a bit of code duplication.
2020-07-16 11:25:08 +01:00
Richard van der Hoff
90b0cdda42 Add some helper classes for generating ID sequences 2020-07-16 11:25:08 +01:00
Erik Johnston
1f36ff69e8
Move event stream handling out of slave store. (#7491)
This allows us to have the logic on both master and workers, which is necessary to move event persistence off master.

We also combine the instantiation of ID generators from DataStore and slave stores to the base worker stores. This allows us to select which process writes events independently of the master/worker splits.
2020-05-15 16:43:59 +01:00
Erik Johnston
8123b2f909
Add MultiWriterIdGenerator. (#7281)
This will be used to coordinate stream IDs across multiple writers.

Functions as the equivalent of both `StreamIdGenerator` and
`SlavedIdTracker`.
2020-05-04 17:17:45 +01:00
Amber Brown
020add5099
Update black to 19.10b0 (#6304)
* update version of black and also fix the mypy config being overridden
2019-11-01 02:43:24 +11:00
Andrew Morgan
4548d1f87e
Remove unnecessary parentheses around return statements (#5931)
Python will return a tuple whether there are parentheses around the returned values or not.

I'm just sick of my editor complaining about this all over the place :)
2019-08-30 16:28:26 +01:00
Amber Brown
7efd1d87c2 Run black on the rest of the storage module (#4996) 2019-04-03 10:07:29 +01:00
Amber Brown
49af402019 run isort 2018-07-09 16:09:20 +10:00
Richard van der Hoff
29ed09e80a Fix assertion to stop transaction queue getting wedged
... and update some docstrings to correctly reflect the types being used.

get_new_device_msgs_for_remote can return a long under some circumstances,
which was being stored in last_device_list_stream_id_by_dest, and was then
upsetting things on the next loop.
2017-03-15 12:16:55 +00:00
Mark Haines
ceb599e789 Add tests for redactions 2016-04-07 16:52:07 +01:00
Mark Haines
9bc5b4c663 Assert that the step != 0 2016-04-01 15:08:20 +01:00
Mark Haines
35b5c4ba1b use google style doc strings 2016-04-01 15:07:01 +01:00
Mark Haines
a2866e2e6a Rename direction to step, apply checks consistently 2016-04-01 13:50:54 +01:00
Mark Haines
e36bfbab38 Use a stream id generator for backfilled ids 2016-04-01 13:29:05 +01:00
Mark Haines
b6e8420aee Add replication stream for pushers 2016-03-15 17:33:10 +00:00
Erik Johnston
158a322e82 Ensure integer is an integer 2016-03-09 10:20:48 +00:00
Mark Haines
a1cf9e3bf3 Add a stream for push rule updates 2016-03-01 18:16:37 +00:00
Mark Haines
54172924c8 Load the current id in the IdGenerator constructor
Rather than loading them lazily. This allows us to remove all
the yield statements and spurious arguments for the get_next
methods.

It also allows us to replace all instances of get_next_txn with
get_next since get_next no longer needs to access the db.
2016-03-01 14:32:56 +00:00
Erik Johnston
42109a62a4 Remove unused param from get_max_token 2016-02-18 16:37:28 +00:00
Erik Johnston
e5999bfb1a Initial cut 2016-02-17 15:40:50 +00:00
Erik Johnston
87f9477b10 Add a Homeserver.setup method.
This is for setting up dependencies that require work on startup. This
is useful for the DataStore that wants to read a bunch from the database
before initiliazing.
2016-01-26 15:51:06 +00:00
Matthew Hodgson
6c28ac260c copyrights 2016-01-07 04:26:29 +00:00
Erik Johnston
b6d4a4c6d8 Merge pull request #199 from matrix-org/erikj/receipts
Implement read receipts.
2015-07-16 18:18:36 +01:00
Erik Johnston
80a61330ee Add basic storage functions for handling of receipts 2015-07-01 17:19:12 +01:00
Erik Johnston
5130d80d79 Add bulk insert events API 2015-06-25 17:29:34 +01:00
Mark Haines
5002056b16 SYN-377: Make sure that the StreamIdGenerator.get_next.__exit__ is called from the main thread after the transaction completes, not from database thread before the transaction completes. 2015-05-12 11:20:40 +01:00
Erik Johnston
0ade2712d1 Typo 2015-04-29 19:17:25 +01:00
Erik Johnston
50f96f256f Also remove yield from within lock in the other generator 2015-04-29 19:17:00 +01:00
Erik Johnston
d2d61a8288 Fix deadlock in id_generators. No idea why this was an actual deadlock. 2015-04-29 19:15:23 +01:00
Erik Johnston
8558e1ec73 Make get_max_token into inlineCallbacks so that the lock works. 2015-04-27 15:19:44 +01:00