In this commit, we extend the wtclient.Client interface with the
following methods:
* AddTower
* RemoveTower
* RegisteredTowers
* LookupTower
* Stats
Care has been taken to ensure that any in-memory state updates are
_only_ performed after a successful database update.
These methods are currently unused, but they serve as a dependency for
the upcoming WatchtowerClient RPC subserver.
These operations are currently unused, but will be integrated into the
TowerClient at a later point as future preparation for the
WatchtowerClient RPC subserver, which will allow users to add, remove,
and list the watchtowers currntly in use.
This currently takes O(N) time as there does not exist an index of
active client sessions for each watchtower within the client's database.
This index is likely to be added in the future.
This commit modifies the client's filtering when selecting from existing
sessions. The new logic compares the configured TxPolicy with the
TxPolicy of the candidate sessions, which has the effect of ignoring
operational parameters such as MaxUpdates. Prior, changing MaxUpdates
would cause the client to request a new session even if it had perfectly
good slots available in a policy with an equal TxPolicy.
This commit splits out the parameters that shape the justice transaction
into their own struct, which then embedded within the overarching
wtpolicy.Policy which may have additional parameters describing
operation of the session.
This is done as a preliminary step to support comparison of sessions
based on matching TxPolicy configurations. This prevents otherwise
identical Policies from being counted as different if operational
parameters like MaxUpdates differ, even if it has no material difference
to the justice transaction.
This commit fixes a bug that would cause us to request more sessions
that needed from the session negotiator. With the current stat ticker,
we'd ask the negotiator for a new session every 30s if session
session negotiation did not return before printing the stats. Now we'll
avoid requesting to sessions by jumping back into the select loop.
Now that the committed and acked updates are persisted across restarts,
we will use them to filter out duplicate commit heights presented by the
client.
This commit replaces the map-based CommittedUpdates field with a slice.
When reading from disk, these will already be sorted by bbolt, so the
client restore the updates as presented without needing to sort them
first.
Since the key in the map variant was the sequence number, we refactor
the CommittedUpdate struct to have a sequence number and an embedded
CommittedUpdateBody (which is equivalent to the old CommittedUpdate).
The database is then expected to populate the sequence number from the
key on disk.
Since the sequence number is now directly integrated in the
CommittedUpdate struct, this allow allows us to remove the now redundant
seqNum argument from CommitUpdate.
