In this commit, we attempt to fix an issue that may lead to force closes due
to small value HTLCs. The sweeper has built in a "negative yield" heuristic
where it won't sweep something that'll result in paying more fees than the
HTLC amount. However for HTLCs, we want to always sweep them, as we don't
cancel back the HTLCs before the outgoing contract is fully resolved.
In the future, we'll start to make more uneconomical decisions about if we
should go to chain at all for small value HTLCs, and also do things like
cancel back early if the HTLC is small and we think we might be contested by
chain fees.
Previously when a block spend is found for the outpoint, our htlc
timeout resolver will do a checkpoint, which implicitly creates a db
record if there isn't one. Now, if the spend is found in mempool,
the resolver will be deleted once the contract is resolved. Later on
when the spend is found in the block again, the resolver will be created
again, but never gets resolved this time.
This commit extends the current htlc timeout resolver to also watch for
preimage spend in mempool for a full node backend.
If mempool enabled, the resolver will watch the spend of the htlc output
in mempool and blocks **concurrently**, as if they are independent.
Ideally, a transaction will first appear in mempool then in a block.
However, there's no guarantee it will appear in **our** mempool since
there's no global mempool, thus we need to watch the spend in two places
in case it doesn't go through our mempool.
The current design favors the spend event found in blocks, that is, when
the tx is confirmed, we'd abort the monitoring and conitnue since the
outpoint cannot be double spent and re-appear in mempool again. This is
not true in the rare case of reorg, and we will handle reorg seperately.
This commit adds a new build tag `integration` and removes the old tag
`rpctest` for clarity. Multiple unnecessary usages of `build !rpctest`
is also removed.
Now that we have the new package `lnd/channeldb/models` we can invert the
depenency between `channeldb` and `invoices`.
- Move all the invoice related types and errors to the
`invoices` package.
- Ensure that all the packages dealing with invoices use the types and
interfaces defined in the `invoices` package.
- Implement the InvoiceDB interface (defined in `lnd/invoices`) in
channeldb.
- Add new mock for InterfaceDB.
- `InvoiceRegistery` tests are now in its own subpacakge (they need to
import both invoices & channeldb). This is temporary until we can
decouple them.
Add a new subpackage to `lnd/channeldb` to hold some of the types that
are used in the package itself and in other packages that should not
depend on `channeldb`.
Only create the sweep tx when the height has reached, otherwise we will
get a broadcast error. Previously it wasn't caught by itest due to the
blocks were mined very fast, so the lock time was always reached.
This introduces a BigSize migration that is used to expand the width
of the ChannelStatus and ChannelType fields. Three channel "types"
are added - ZeroConfBit, ScidAliasChanBit, and ScidAliasFeatureBit.
ScidAliasChanBit denotes that the scid-alias channel type was
negotiated for the channel. ScidAliasFeatureBit denotes that the
scid-alias feature bit was negotiated during the *lifetime* of the
channel. Several helper functions on the OpenChannel struct are
exposed to aid callers from different packages.
The RefreshShortChanID has been renamed to Refresh.
A new function BroadcastHeight is used to guard access to the
mutable FundingBroadcastHeight member. This prevents data races.
In this commit, we consolidate the _lease specific_ logic for the
success and timeout HTLC resolvers. We do this with the addition of a
new struct which is then composed via struct embedding with the two
existing structs. This fixes a flake in the integration tests by
ensuring the height is set up front, rather than eventually once the
height matches the lock time.
This prevents a panic where an insert is attempted on a nil map via
updateActiveHTLCs. This panic would occur if the channel arbitrator
was in a buggy state, possibly introduced by power loss or via
SIGKILL.
The main idea is that NotifyContractUpdate adds the ContractUpdate to
a map called unmerged. It is populated in Start by shallow-copying the
activeHTLCs map values (htlcSet). The htlcSets underlying maps are not
copied, and so unmerged will just contain pointers to them. This should
be fine since unmerged will not modify them. At the call-sites of
activeHTLCs, it is updated to include the unmerged sets. This happens
with a mutex and should not cause any data race, even though it is
copying the underlying map pointers. No persistence should be
necessary since on restart, activeHTLCs and unmerged will just be
populated again.
This commit changes the `NewBreachRetribution` to use the new revocation
log format, while maintaining the compatibilty to use an older
revocation log format. Unit tests have been added to make sure a breach
retribution can be created in both log formats.
This also means the watch tower needs to pass the relevant commit tx at
its backup height when creating the breach retribution during backing
up. This is achieved by recording the current remote commitment state
before advancing the remote commitment chain.
This commit was previously split into the following parts to ease
review:
- 2d746f68: replace imports
- 4008f0fd: use ecdsa.Signature
- 849e33d1: remove btcec.S256()
- b8f6ebbd: use v2 library correctly
- fa80bca9: bump go modules
This also changes the chain_watcher and breacharbiter handoff. The
new logic ensures that the channel is only marked as pending closed
when the channel arbitrator has persisted the resolutions and commit
set.
