mirror of
https://mau.dev/maunium/synapse.git
synced 2024-12-14 03:53:47 +01:00
Reduce log spam when running multiple event persisters (#12610)
This commit is contained in:
parent
2d74a8c178
commit
c0379d6e5b
3 changed files with 17 additions and 2 deletions
1
changelog.d/12610.misc
Normal file
1
changelog.d/12610.misc
Normal file
|
@ -0,0 +1 @@
|
|||
Reduce log spam when running multiple event persisters.
|
|
@ -537,7 +537,7 @@ class ReplicationCommandHandler:
|
|||
# Ignore POSITION that are just our own echoes
|
||||
return
|
||||
|
||||
logger.info("Handling '%s %s'", cmd.NAME, cmd.to_line())
|
||||
logger.debug("Handling '%s %s'", cmd.NAME, cmd.to_line())
|
||||
|
||||
self._add_command_to_stream_queue(conn, cmd)
|
||||
|
||||
|
@ -567,6 +567,11 @@ class ReplicationCommandHandler:
|
|||
# between then and now.
|
||||
missing_updates = cmd.prev_token != current_token
|
||||
while missing_updates:
|
||||
# Note: There may very well not be any new updates, but we check to
|
||||
# make sure. This can particularly happen for the event stream where
|
||||
# event persisters continuously send `POSITION`. See `resource.py`
|
||||
# for why this can happen.
|
||||
|
||||
logger.info(
|
||||
"Fetching replication rows for '%s' between %i and %i",
|
||||
stream_name,
|
||||
|
@ -590,7 +595,7 @@ class ReplicationCommandHandler:
|
|||
[stream.parse_row(row) for row in rows],
|
||||
)
|
||||
|
||||
logger.info("Caught up with stream '%s' to %i", stream_name, cmd.new_token)
|
||||
logger.info("Caught up with stream '%s' to %i", stream_name, cmd.new_token)
|
||||
|
||||
# We've now caught up to position sent to us, notify handler.
|
||||
await self._replication_data_handler.on_position(
|
||||
|
|
|
@ -204,6 +204,15 @@ class ReplicationStreamer:
|
|||
# turns out that e.g. account data streams share
|
||||
# their "current token" with each other, meaning
|
||||
# that it is *not* safe to send a POSITION.
|
||||
|
||||
# Note: `last_token` may not *actually* be the
|
||||
# last token we sent out in a RDATA or POSITION.
|
||||
# This can happen if we sent out an RDATA for
|
||||
# position X when our current token was say X+1.
|
||||
# Other workers will see RDATA for X and then a
|
||||
# POSITION with last token of X+1, which will
|
||||
# cause them to check if there were any missing
|
||||
# updates between X and X+1.
|
||||
logger.info(
|
||||
"Sending position: %s -> %s",
|
||||
stream.NAME,
|
||||
|
|
Loading…
Reference in a new issue