In order to be consistent with other sub systems an error is now
returned from the Stop functions.
This also allows writing a generic cleanup mechanism to stop all
sub systems in case of a failure.
Previously, we would always allow dependent jobs to be processed,
regardless of the result of its parent job's validation. This isn't
correct, as a parent job contains actions necessary to successfully
process a dependent job. A prime example of this can be found within the
AuthenticatedGossiper, where an incoming channel announcement and update
are both processed, but if the channel announcement job fails to
complete, then the gossiper is unable to properly validate the update.
This commit aims to address this by preventing the dependent jobs to
run.
The recently added gossip throttling was shown to be too aggressive,
especially with our auto channel enable/disable signaling. We switch to
a token bucket based system instead as it's based on time, rather than a
block height which isn't constantly updated at a given rate.
Since the batch interval can potentially be long, adding local updates
to the graph could be slow. This would slow down operations like adding
our own channel update and announcements during the funding process, and
updating edge policies for local channels.
Now we instead check whether the update is remote or not, and only for
remote updates use the SchedulerOption to lazily add them to the graph.
We do this instead of using the source of the AnnounceSignatures
message, as we filter out the source when broadcasting any
announcements, leading to the remote node not receiving our channel
update. Note that this is done more for the sake of correctness and to
address a flake within the integration tests, as channel updates are
sent directly and reliably to channel counterparts.
A pinned syncer is an ActiveSyncer that is configured to always remain
active for the lifetime of the connection. Pinned syncers do not count
towards the total NumActiveSyncer count, which are rotated periodically.
This features allows nodes to more tightly synchronize their routing
tables by ensuring they are always receiving gossip from distinguished
subset of peers.
We do this instead of using the source of the AnnounceSignatures
message, as we filter out the source when broadcasting any
announcements, leading to the remote node not receiving our channel
update. Note that this is done more for the sake of correctness and to
address a flake within the integration tests, as channel updates are
sent directly and reliably to channel counterparts.
As similarly done with premature channel announcements, we'll no longer
allow premature channel updates to be rebroadcast once mature. This is
no longer necessary as channel announcements that we're not aware of are
usually broadcast to us with their accompanying channel updates.
In this commit, we add a new option to toggle gossip rate limiting. This
new option can be useful in contexts that require near instant
propagation of gossip messages like integration tests.
This change was largely motivated by an increase in high disk usage as a
result of channel update spam. With an in memory graph, this would've
gone mostly undetected except for the increased bandwidth usage, which
this doesn't aim to solve yet. To minimize the effects to disks, we
begin to rate limit channel updates in two ways. Keep alive updates,
those which only increase their timestamps to signal liveliness, are now
limited to one per lnd's rebroadcast interval (current default of 24H).
Non keep alive updates are now limited to one per block per direction.
This allows for a 1000 different validation operations to proceed
concurrently. Now that we are batching operations at the db level, the
average number of outstanding requests will be higher since the commit
latency has increased. To compensate, we allow for more outstanding
requests to keep the gossiper busy while batches are constructed.
This reworks the locking behavior of the Gossiper so that a race
condition on channel updates and block notifications doesn't cause any
loss of messages.
This fixes an issue that manifested mostly as flakes on itests during
WaitForNetworkChannelOpen calls.
The previous behavior allowed ChannelUpdates to be missed if they
happened concurrently to block notifications. The
processNetworkAnnoucement call would check for the current block height,
then lock the gossiper and add the msg to the prematureAnnoucements
list. New blocks would trigger an update to the current block height
then a lock and check of the aforementioned list.
However, specially during itests it could happen that the missing lock
before checking the height could case a race condition if the following
sequence of events happened:
- A new ChannelUpdate message was received and started processing on a
separate goroutine
- The isPremature() call was made and verified that the ChannelUpdate
was in fact premature
- The goroutine was scheduled out
- A new block started processing in the gossiper. It updated the block
height, asked and was granted the lock for the gossiper and verified
there was zero premature announcements. The lock was released.
- The goroutine processing the ChannelUpdate asked for the gossiper lock
and was granted it. It added the ChannelUpdate in the
prematureAnnoucements list. This can never be processed now.
The way to fix this behavior is to ensure that both isPremature checks
done inside processNetworkAnnoucement and best block updates are made
inside the same critical section (i.e. while holding the same lock) so
that they can't both check and update the prematureAnnoucements list
concurrently.
The policy update logic that resided part in the gossiper and
part in the rpc server is extracted into its own object.
This prepares for additional validation logic to be added for policy
updates that would otherwise make the gossiper heavier.
It is also a small first step towards separation of our own channel data
from the rest of the graph.
As a preparation for making the gossiper less responsible for validating
and supplementing local channel policy updates, this commits moves the
on-the-fly max htlc migration up the call tree. The plan for a follow up
commit is to move it out of the gossiper completely for local channel
updates, so that we don't need to return a list of final applied policies
anymore.
In this commit, we fix a bug where if a user updates a forwarding policy to be
zero, the update will be applied to the policy correctly on-disk, but not
in-memory.
We solve this issue by having the gossiper return the list of on-disk updated
policies and passing these policies to the switch, so the switch can assume
that zero-valued fields are intentional and not just uninitialized.
There's no need to broadcast these as we assume that online nodes have
already received them. For nodes that were offline, they should receive
them as part of their initial graph sync.
We do this to ensure the node announcement propagates to our channel
counterparty. At times, the node announcement does not propagate to them
when opening our first channel due to a race condition between
IsPublicNode and processing announcement signatures. This isn't
necessary for channel updates and announcement signatures as we send
those to our channel counterparty directly through the reliable sender.
