We plan to later on add an option for a remote graph source which will
be managed from the ChannelGraph. In such a set-up, a node would rely on
the remote graph source for graph updates instead of from gossip sync.
In this scenario, however, our topology subscription logic should still
notify clients of all updates and so it makes more sense to have the
logic as part of the ChannelGraph so that we can send updates we receive
from the remote graph.
The test as it stands today does not make sense as it adds a
Partial/Shell node to the graph via AddLightningNode which will never
happen since this is only ever triggered by the gossiper which only
calls the method with a full node announcement. Shell/Partial nodes are
only ever added via AddChannelEdge which will insert a partial node if
we are adding a channel edge which has node pub keys that we dont have a
node entry for. So we adjust the test to use this more accurate flow.
We do this in preparation for moving channel cache population logic out
of the constructor and into the Start method. We also will later on
(when topology subscription is moved to the ChannelGraph), have a
goroutine that will need to be kicked off and stopped.
Here, we move the graph cache writes for AddLightningNode,
DeleteLightningNode, AddChannelEdge and MarkEdgeLive to the
ChannelGraph. Since these are writes, the cache is only updated if the
DB write is successful.
This commit moves the graph cache checks for FetchNodeFeatures,
ForEachNodeDirectedChannel, GraphSession and ForEachNodeCached from the
KVStore to the ChannelGraph. Since the ChannelGraph is currently just a
pass-through for any of the KVStore methods, all that needs to be done
for calls to go via the ChannelGraph instead directly to the KVStore is
for the ChannelGraph to go and implement those methods.
In this commit, we let the ChannelGraph be responsible for populating
the graphCache and then passing it to the KVStore. This is a first step
in moving the graphCache completely out of the KVStore layer.
And use this struct to pass NewChannelGraph anything it needs to be able
to init the KVStore that it houses. This will allow us to add
ChannelGraph specific options.
Since we have renamed a file housing some very old code, the linter has
now run on all this code for the first time. So we gotta do some
clean-up work here to make it happy.
In this commit, we rename the existing ChannelGraph struct to KVStore to
better reflect its responsibilities as the CRUD layer. We then introduce
a new ChannelGraph struct which will eventually be the layer above the
CRUD layer in which we will handle cacheing and topology subscriptions.
For now, however, it houses only the KVStore. This means that all calls
to the KVStore will now go through this layer of indirection first. This
will allow us to slowly move the graph Cache management out of the
KVStore and into the new ChannelGraph layer.
We introduce the new ChannelGraph and rename the old one in the same
commit so that all existing call-sites don't need to change at all :)
Rename it to kv_store.go so that we can re-use the graph.go file name
later on. We will use it to house the _new_ ChannelGraph when the
existing ChannelGraph is renamed to more clearly reflect its
responsibilities as the CRUD layer.
This commit adds a test to demonstrate that if we receive two identical
updates (which can happen if we get the same update from two peers in
quick succession), then our rate limiting logic will be hit early as
both updates might be counted towards the rate limit. This will be fixed
in an upcoming commit.
Before this commit, it was possible for a request to be sent on the
`chanWatchRequests` channel in `WaitForChannelPolicyUpdate` and then for
the `ticker.C` case to select _before_ the `eventChan` select gets
triggered when the `topologyWatcher` closes the `eventChan` in its call
to `handlePolicyUpdateWatchRequest`. This could lead to a "close of a
closed channel" panic.
To fix this, this commit ensures that we only move on to the next
iteration of the select statement in `WaitForChannelPolicyUpdate` once
the request sent on `chanWatchRequests` has been fully handled.
The exposed AddNode, AddEdge and UpdateEdge methods of the Builder are
currently synchronous since even though they pass messages to the
network handler which spins off the handling in a goroutine, the public
methods still wait for a response from the handling before returning.
The only part that is actually done asynchronously is the topology
notifications.
We previously tried to simplify things in [this
commit](https://github.com/lightningnetwork/lnd/pull/9476/commits/d757b3bcfc77b0fbb543ef5db800e6c6b058930f)
but we soon realised that there was a reason for sending the messages to
the central/synchronous network handler first: it was to ensure
consistency for topology clients: ie, the ordering between when there is
a new topology client or if it is cancelled needs to be consistent and
handled synchronously with new network updates. So for example, if a new
update comes in right after a topology client cancels its subscription,
then it should _not_ be notified. Similariy for new subscriptions. So
this commit was reverted soon after.
We can, however, still simplify things as is done in this commit by
noting that _only topology subscriptions and notifications_ need to be
handled separately. The actual network updates do not need to. So that
is what is done here.
This refactor will make moving the topology subscription logic to a new
subsystem later on much easier.
This commit restricts the graph CRUD interface such that one can only
add a proof to a channel announcement and not update any other fields.
This pattern is better suited for SQL land too.
This commit fixes a bug that would check that the passed `edge` argument
of UpdateChannelEdge is nil but it should actually be checking if the
`edges` bucket is nil.
In this commit, we use the available kvdb `View` method directly for
starting a graph session instead of manually creating and commiting the
transaction. Note this changes behaviour since failed tx create/commit
will now error instead of just log.
In this commit, we add a `GraphSession` method to the `ChannelGraph`.
This method provides a caller with access to a `NodeTraverser`. This is
used by pathfinding to create a graph "session" overwhich to perform a
set of queries for a pathfinding attempt. With this refactor, we hide
details such as DB transaction creation and transaction commits from the
caller. So with this, pathfinding does not need to remember to "close
the graph session". With this commit, the `graphsession` package may be
completely removed.
In preparation for the next commit where we will start hiding underlying
graph details such as that a graph session needs to be "closed" after
pathfinding is done with it, we refactor things here so that the main
pathfinding logic is done in a call-back.
The `graphsession.NewRoutingGraph` method was used to create a
RoutingGraph instance with no consistent read transaction across calls.
But now that the ChannelGraph directly implements this, we can remove
The NewRoutingGraph method.
In preparation for having the ChannelGraph directly implement the
`routing.Graph` interface, we rename the `ForEachNodeChannel` method to
`ForEachNodeDirectedChannel` since the ChannelGraph already uses the
`ForEachNodeChannel` name and the new name is more appropriate since the
ChannelGraph currently has a `ForEachNodeDirectedChannelTx` method which
passes the same DirectedChannel type to the given call-back.
Add the `Tx` suffix to both ForEachNodeDirectedChannelTx and
FetchNodeFeatures temporarily so that we free up the original names for
other use. The renamed methods will be removed or unexported in an
upcoming commit. The aim is to have no exported methods on the
ChannelGraph that accept a kvdb.RTx as a parameter.
For consistency in the graphsessoin.graph interface, we let the
FetchNodeFeatures method take a read transaction just like the
ForEachNodeDirectedChannel. This is nice because then all calls in the
same pathfinding transaction use the same read transaction.
This commit is a pure refactor. We move the transaction validation
(existence, spentness, correctness) from the `graph.Builder` to the
gossiper since this is where all protocol level checks should happen.
All tests involved are also updated/moved.
As we move the funding transaction validation logic out of the builder
and into the gossiper, we want to ensure that the behaviour stays
consistent with what we have today. So we should aquire this lock before
performing any expensive checks such as building the funding tx or
valdating it.
In preparation for an upcoming commit which will move all channel
funding tx validation to the gossiper, we first move the helper method
which helps build the expected funding transaction script based on the
fields in the channel announcement. We will still want this script later
on in the builder for updating the ChainView though, and so we pass this
field along with the ChannelEdgeInfo. With this change, we can remove
the TapscriptRoot field from the ChannelEdgeInfo since the only reason
it was there was so that the builder could reconstruct the full funding
script.
Here, we add a new fundingTxOption modifier which will configure how we
set-up expected calls to the mock Chain once we have moved funding tx
logic to the gossiper. Note that in this commit, these modifiers don't
yet do anything.
This is in preparation for the commit where we move across all the
funding tx validation so that we can test that we are correctly updating
the zombie index.
The `graph.Builder`'s `addZombieEdge` method is currently called during
funding transaction validation for the case where the funding tx is not
found. In preparation for moving this code to the gossiper, we export
the method and add it to the ChannelGraphSource interface so that the
gossiper will be able to call it later on.
Sometimes only the "Send coverage" step of a CI job will fail. This
commit turns this step into a "best effort" step instead so that it
does not block a CI job from passing.
It can, for example, often happen that a single job is re-run from the
GH UI, it then passes but the "Send coverage" step fails due to the
"Can't add a job to a build that is already closed." error meaning that
the only way to get the CI step to pass is to re-push and retrigger a
full CI run.
With the previous commit, the AddNode method was removed and since that
was the only method making use of the ForEachChannel on the
GraphCacheNode interface, we can remove that method. Since the only two
methods left just expose the node's pub key and features, it really is
not required anymore and so the entire thing can be removed along with
the implementation of it.
The AddNode method on the GraphCache calls `AddNodeFeatures` underneath
and then iterates through all the node's persisted channels and adds
them to the cache too via `AddChannel`.
This is, however, not required since at the time the cache is populated
in `NewChannelGraph`, the cache is populated will all persisted nodes
and all persisted channels. Then, once any new channels come in, via
`AddChannelEdge`, they are added to the cache via AddChannel. If any new
nodes come in via `AddLightningNode`, then currently the cache's AddNode
method is called which the both adds the node and again iterates through
all persisted channels and re-adds them to the cache. This is definitely
redundent since the initial cache population and updates via
AddChannelEdge should keep the cache fresh in terms of channels.
So we remove this for 2 reasons: 1) to remove the redundent DB calls and
2) this requires a kvdb.RTx to be passed in to the GraphCache calls
which will make it hard to extract the cache out of the CRUD layer
and be used more generally.
The AddNode method made sense when the cache was first added in the
code-base
[here](https://github.com/lightningnetwork/lnd/commit/369c09be6152b76a24915050a8a5fe6bccf2b8f0#diff-ae36bdb6670644d20c4e43f3a0ed47f71886c2bcdf3cc2937de24315da5dc072R213)
since then during graph cache population, nodes and channels would be
added to the cache in a single DB transaction. This was, however,
changed [later
on](https://github.com/lightningnetwork/lnd/commit/352008a0c22ffd9e97b9481ffa44dbc50fca2ddc)
to be done in 2 separate DB calls for efficiency reasons.
Define a new GraphSource interface that describes the access required by
the autopilot server. Let its constructor take this interface instead of
a raw pointer to the graphdb.ChannelGraph.
In this commit, a new NodeRTx interface is added which represents
consistent access to a persisted models.LightningNode. The
ForEachChannel method of the interface gives the caller access to the
node's channels under the same read transaction (if any) that was used
to fetch the node in the first place. The FetchNode method returns
another NodeRTx which again will have the same underlying read
transaction.
The main point of this interface is to provide this consistent access
without needing to expose the `kvdb.RTx` type as a method parameter.
This will then make it much easier in future to add new implementations
of this interface that are backed by other databases (or RPC
connections) where the `kvdb.RTx` type does not apply.
We will make use of the new interface in the `autopilot` package in
upcoming commits in order to remove the `autopilot`'s dependence on the
pointer to the `*graphdb.ChannelGraph` which it has today.
Introduce a new type for testing code so the main databaseChannelGraph
type does not need to make various write calls to the
`graphdb.ChannelGraph` but the new testDBGraph type still can for tests.
This is a pure code move commit where we move any code that is only ever
used by tests to test files. Many of the calls to the
graphdb.ChannelGraph pointer are only coming from tests code.
In preparation for adding more modifiers. We want to later add a
modifier that will tweak the errors returned by the mock chain once
funding transaction validation has been moved to the gossiper.
This is in preparation for moving the funding transaction validation
code to the gossiper from the graph.Builder since then the gossiper will
start making GetBlockHash/GetBlock and GetUtxo calls.
Convert a bunch of the helper functions to instead be methods on the
testCtx type. This is in preparation for adding a mockChain to the
testCtx that these helpers can then use to add blocks and utxos to.
See `notifications_test.go` for an idea of what we are trying to emulate
here. Once the funding tx code has moved to the gossiper, then the logic
in `notifications_test.go` will be removed.
In preparation for moving funding transaction validiation from the
Builder to the Gossiper in later commit, we first convert these graph
Error Codes to normal error variables. This will help make the later
commit a pure code move.
The point of the `graph.Builder`'s `networkHandler` goroutine is to
ensure that certain requests are handled in a synchronous fashion.
However, any requests received on the `networkUpdates` channel, are
currently immediately handled in a goroutine which calls
`handleNetworkUpdate` which calls `processUpdate` before doing topology
notifications. In other words, there is no reason for these
`networkUpdates` to be handled in the `networkHandler` since they are
always handled asynchronously anyways. This design is most likely due to
the fact that originally the gossiper and graph builder code lived in
the same system and so the pattern was copied across.
So in this commit, we just remove the complexity. The only part we need
to spin off in a goroutine is the topology notifications.
In this commit, we remove the `processUpdate` method which handles each
announement type (node, channel, channel update) in a separate switch
case. Each of these cases currently has a non-trivial amount of code.
This commit creates separate methods for each message type we want to
handle instead. This removes a level of indentation and will make things
easier to review when we start editing the code for each handler.
The `netann` package is a more appropriate place for this code to live.
Also, once the funding transaction code is moved out of the
`graph.Builder`, then no `lnwire` validation will occur in the `graph`
package.
This commit updates the fn dep to the version containing the updates to
the ContextGuard implementation. Only the htlcswitch/link uses the guard
at the moment so this is updated to make use of the new implementation.
This bump includes a fix which prevents attribute value quoting if the
value string contains a newline character. This is so that if we call
spew.DumpS(), the output will stay nicely formatted.
The update also includes a couple more Hex helpers which we can make use
of now.
In this commit, the ContextGuard struct is re-worked such that the
context that its new main WithCtx method provides is cancelled in sync
with a parent context being cancelled or with it's quit channel being
cancelled. Tests are added to assert the behaviour. In order for the
close of the quit channel to be consistent with the cancelling of the
derived context, the quit channel _must_ be contained internal to the
ContextGuard so that callers are only able to close the channel via the
exposed Quit method which will then take care to first cancel any
derived context that depend on the quit channel before returning.
Elle Mouton