When a `chain::Watch` `ChannelMonitor` update method is called, the
user has three options:
(a) persist the monitor update immediately and return success,
(b) fail to persist the monitor update immediately and return
failure,
(c) return a flag indicating the monitor update is in progress and
will complete in the future.
(c) is rather harmless, and in some deployments should be expected
to be the return value for all monitor update calls, but currently
requires returning `Err(ChannelMonitorUpdateErr::TemporaryFailure)`
which isn't very descriptive and sounds scarier than it is.
Instead, here, we change the return type used to be a single enum
(rather than a Result) and rename `TemporaryFailure`
`UpdateInProgress`.
As we move towards specify supported/required feature bits in the
module(s) where they are supported, the global `known` feature set
constructors no longer make sense.
Here we stop relying on the `known` method in the
`lightning-background-processor` and `lightning-persister` crate
tests.
If a user restores from a backup that they know is stale, they'd
like to force-close all of their channels (or at least the ones
they know are stale) *without* broadcasting the latest state,
asking their peers to do so instead. This simply adds methods to do
so, renaming the existing `force_close_channel` and
`force_close_all_channels` methods to disambiguate further.
We generally make no effort to ensure all writes are buffered in
lower-level objects, so wrapping write calls in `BufWriter` may
substantially improve performance in some cases. This is especially
important now that we block the sample node exit until the
`NetworkGraph` has been written out, which includes many small-ish
writes.
With this change, shutdown of the sample node on a relatively
underpowered device went from 15-30 seconds of CPU time to a second
or two, plus IO sync time.
Instead of creating a separate trait for persisting NetworkGraph, use and rename the existing ChannelManagerPersister to handle them both. persist_graph is then called on removal of stale channels and on exit.
If we are in the middle of persisting an update to a
`ChannelMonitor` when we shutdown (or crash), we'll start up with
a .tmp file lying around. We should ignore it, as failure to
return from the update call should have prevented the
`ChannelManager` from taking any irrevocable action based on the
update.
We're somewhat protected from any filesystem inconsistency behavior
as the `ChannelManager` will refuse to load if we're outright
missing `ChannelMonitor`s.
Fixes#1330.
`cargo bench` sets `cfg(test)`, causing us to hit some test-only
code in the router when benchmarking, throwing off our benchmarks
substantially. Here we swap from the `unstable` feature to a more
clearly internal feature (`_bench_unstable`) and also checking for
it when enabling test-only code.
This resolves several user complaints (and issues in the sample
node) where startup is substantially delayed as we're always
waiting for the chain data to sync.
Further, in an upcoming PR, we'll be reloading pending payments
from ChannelMonitors on restart, at which point we'll need the
change here which avoids handling events until after the user
has confirmed the `ChannelMonitor` has been persisted to disk.
It will avoid a race where we
* send a payment/HTLC (persisting the monitor to disk with the
HTLC pending),
* force-close the channel, removing the channel entry from the
ChannelManager entirely,
* persist the ChannelManager,
* connect a block which contains a fulfill of the HTLC, generating
a claim event,
* handle the claim event while the `ChannelMonitor` is being
persisted,
* persist the ChannelManager (before the CHannelMonitor is
persisted fully),
* restart, reloading the HTLC as a pending payment in the
ChannelManager, which now has no references to it except from
the ChannelMonitor which still has the pending HTLC,
* replay the block connection, generating a duplicate PaymentSent
event.
In the next commit, we'll be originating monitor updates both from
the ChainMonitor and from the ChannelManager, making simple
sequential update IDs impossible.
Further, the existing async monitor update API was somewhat hard to
work with - instead of being able to generate monitor_updated
callbacks whenever a persistence process finishes, you had to
ensure you only did so at least once all previous updates had also
been persisted.
Here we eat the complexity for the user by moving to an opaque
type for monitor updates, tracking which updates are in-flight for
the user and only generating monitor-persisted events once all
pending updates have been committed.
When we detect a channel `is_shutdown()` or call on it
`force_shutdown()`, we notify the user with a Event::ChannelClosed
informing about the id and closure reason.
std::io::ErrorKind is a `#[non_exhaustive]` enum as more specific
error types are to be added in the future. It was unclear in the
docs until very recently, however, that this is to be done by
re-defining `ErrorKind::Other` errors to new enum variants. Thus,
our tests which check explicitly for `ErrorKind::Other` as a
result of trying to access a directory as a file were incorrect.
Sadly, these generated no meaningful feedback from rustc at all,
except that they're suddenly failing in rustc beta!
After some back-and-forth, it seems rustc is moving forward
breaking existing code in future versions, so we move to the
"correct" check here, which is to check the raw IO error.
See rust-lang/rust#86442 and rust-lang/rust#85746 for more info.
Like the payment_secret parameter, this paramter has been the source
of much confusion, so we just drop it.
Users should prefer to do this check when registering the payment
secret instead of at claim-time.
There's little reason for the HashMap - the ChannelMonitors are
already unique (enforced by file names), and the eventual HashMap
that users need when deserializing the `ChannelManager` is a
slightly different form (it requires no BlockHash entry).
