If a subscriber arrives while the pubsub service is shutting down, the existing
code will return a nil subscription without error. With unlucky timing, this
may lead to a nil indirection panic in the RPC service.
To avoid that problem, make sure that when a subscription fails for this
reason, we report a non-nil error so that the client will detect it and give up
gracefully.
A workaround for #6729. Add parameters to control buffer sizes for
event subscription RPC clients. On some networks, buffering causes
clients to be dropped and/or events to be lost.
For additional context, see the discussion on #7188.
- Add experimental_subscription_buffer_size config parameter
- Add experimental_websocket_write_buffer_size config parameter
- Add experimental_close_on_slow_client config parameter
Co-authored-by: M. J. Fromberger <fromberger@interchain.io>
This change backports the PostgreSQL indexing sink, addressing part of #6828.
Development on the main branch has diverged substantially since the v0.34.x
release. It includes package moves, breaking API and protobuf schema changes,
and new APIs, all of which together have a large footprint on the mapping
between the implementation at tip and the v0.34 release branch.
To avoid the need to retrofit all of those improvements, this change works by
injecting the new indexing sink into the existing (v0.34) indexing interfaces
by delegation. This means the backport does _not_ pull in all the newer APIs
for event handling, and thus has minimal impact on existing code written
against the v0.34 package structure.
This change includes the test for the `psql` implementation, and thus updates
some Go module dependencies. Because it does not interact with any other types,
however, I did not add any unit tests to other packages in this change.
Related changes:
* Update module dependencies for psql backport.
* Update test data to be type-compatible with the old protobuf types.
* Add config settings for the PostgreSQL indexer.
* Clean up some linter settings.
* Hook up the psql indexer in the node main.
* version: revert version through ldflag only (#6494)
Add version back to versions, but allow it to be overridden via a ldflag.
Reason:
Many users are not setting the ldflag causing issues with tooling that relies on it (cosmjs)
closes#6488
cc @webmaster128
* revert variable rename
* Update CHANGELOG_PENDING.md
@p4u from vocdoni.io reported that the mempool might behave incorrectly under a
high load. The consequences can range from pauses between blocks to the peers
disconnecting from this node.
My current theory is that the flowrate lib we're using to control flow
(multiplex over a single TCP connection) was not designed w/ large blobs
(1MB batch of txs) in mind.
I've tried decreasing the Mempool reactor priority, but that did not
have any visible effect. What actually worked is adding a time.Sleep
into mempool.Reactor#broadcastTxRoutine after an each successful send ==
manual control flow of sort.
As a temporary remedy (until the mempool package
is refactored), the max-batch-bytes was disabled. Transactions will be sent
one by one without batching
Closes#5796
When set to true, an invalid transaction will be kept in the cache (this may help some applications to protect against spam).
NOTE: this is a temporary config option. The more correct solution would be to add a TTL to each transaction (i.e. CheckTx may return a TTL in ResponseCheckTx).
Closes: #5751
M. J. Fromberger
Thane Thomson
Callum Waters
mergify[bot]
Tess Rinearson
Marko
Gustavo Chaín
Anton Kaliaev
Erik Grinaker