We were actually using the last commit tx's size, since we were
setting it in lightningd. Instead, hand the min and desired feerates
to closingd, and (as it knows the weight of the closing tx), and have
it start negotiation from there.
This can be significantly less when anchor outputs are enabled: for
example in test_closing.py, the commit tx weight is 1124 Sipa, the
close is 672 Sipa!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Changed: Protocol: Use a more accurate fee for mutual close negotiation.
We set amounts to a list, but then don't use the values except
as a boolean.
Make it a boolean, which should also speed the test up!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This lets us transition (with a few supporting changes) to closingd,
which will happily let them mutual close with us.
We already handle the case where this mutual close is redundant (for
packet loss), so this is easy.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: Protocol: We will now reestablish and negotiate mutual close on channels we've already closed (great if peer has lost their database).
In particular, if one side sees the final CLOSING_SIGNED and the other
doesn't, we won't talk when it reconnects.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It handles all the cases of retransmission, and in the normal case
retransmits shutdown and immediately returns for us to run closingd.
This is actually far simpler and reduces code duplication.
[ Includes fixup to stop warn_unused_result from Christian ]
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: We could get stuck on signature exchange if we needed to retransmit the final revoke_and_ack.
This was turned by at random by CI:
1. Alice has sent shutdown, but it still waiting for revoke_and_ack.
2. Bob has sent and received shutdown, and sent revoke_and_ack,
so it considers it time for signature exchange.
3. Disconnect before Alice received revoke_and_ack.
4. Reconnect, Bob is in closingd, which doesn't rexmit revoke_and_ack.
5. Timeout.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Temporarily rename old getroute to getrouteold (we will remove this).
Changelog-Changed: JSON-RPC: `getroute` is now implemented in a plugin.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Prior to this, sending a v1 address (or, in fact, any random crap!)
would cause the unsupporting node to unilaterally close.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This includes anysegwit and the updated HTLC tiebreak test vector. It
also adds explicit wording for invalid per_commitment_secret (which
nicely matches our code already!).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Users are more upset recently with the cost of unilateral closes
than they are the risk of being cheated. While we complete our
anchor implementation so we can use low fees there, let's
get less aggressive (we already have 34 or 18 blocks to close
in the worst case).
The changes are:
- Commit transactions were "2 CONSERVATIVE" now "6 ECONOMICAL".
- HTLC resolution txs were "3 CONSERVATIVE" now "6 ECONOMICAL".
- Penalty txs were "3 CONSERVATIVE" now "12 ECONOMICAL".
- Normal txs were "4 ECONOMICAL" now "12 ECONOMICAL".
There can be no perfect levels, but we have had understandable
complaints recently about how high our default fee levels are.
Changelog-Changed: Protocol: channel feerates reduced to bitcoind's "6 block ECONOMICAL" rate.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't always get two transactions on line 1019; the comment is
confused (only one penalty tx successfully comes out of l3). So make
sure we get the other transactions when we expect them, and then
make this test more specific.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: lightningd: experimental-shutdown-wrong-funding to allow remote nodes to close incorrectly opened channels.
Changelog-Added: JSON-RPC: close has a new `wrong_funding` option to try to close out unused channels where we messed up the funding tx.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The spec doesn't say to do this, but it makes sense, otherwise
they'll never be able to mutually close the channel.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Since we turned many errors into warnings, we want our tests to fail
when they happen unexpectedly. We make WARNING clear in the strings
we print, too, to help out.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The test was not considering that concurrent sendrawtx of the same tx
is not stable, and either endpoint will submit it first. Now just
checking state transitions and the mempool.
We also make the logic a bit nicer to read. The failure was due to
more than one status message being present if we look at the wrong
time:
```
arr = ['CLOSINGD_SIGEXCHANGE:We agreed on a closing fee of 20334 satoshi for tx:17f1e9d377840edf79d8b6f1ed0faba59bb307463461...9b98', 'CLOSINGD_SIGEXCHANGE:Waiting for another closing fee offer: ours was 20334 satoshi, theirs was 20332 satoshi,'] │
def only_one(arr):
"""Many JSON RPC calls return an array; often we only expect a single entry
"""
> assert len(arr) == 1
E AssertionError
```
We weren't waiting for the transactions to enter the mempool which
could cause all of our fine-tuned block counts to be off. Now just
waiting for the expected number of txs.
For compatibility, we only do this if `allow-deprecated-apis` is false
for now. Otherwise scripts parsing should use `grep -v '^# '` or
start using `-N none`.
Changelog-Added: JSON-RPC: `close` now sends notifications for slow closes (if `allow-deprecated-apis`=false)
Changelog-Deprecated: cli: scripts should filter out '^# ' or use `-N none`, as commands will start returning notifications soon
Fixes: #3925
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're rarely in a hurry here, and bitcoind is aggressive with fees.
You can always spend this output if you really have to, using CPFP.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Changed: Protocol: mutual closing feerate reduced to "slow" to avoid overpaying.
lightning-5 can sometimes see itself sweeping the unilateral output resulting
in this weird line:
```log
HTLC already resolved by SELF when we found preimage
```
Marking spent means if the transaction doesn't confirm for some
reason, the user will need to force a rescan to find the funds. Now
we have timed reservations, reserving for (an additional) 12 hours
should be sufficient.
We also take this opportunity (now we have our own callback path)
to record the tx in the wallet only on success.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Anchor outputs break many assumptions in our tests:
1. Remove some hardcoded numbers in favor of a fee calc, so we only have to
change in one place.
FIXME: This should also be done for elements!
2. Do binary search to get feerate for a given closing fee.
3. Don't assume output #0: anchor outputs perturb them.
4. Don't assume we can make 1ksat channels (anchors cost 660 sats!).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It might have already happened, and anyway, we wait for the actual state
below.
```
# make database snapshot of l2
l2.stop()
l2_db_path = os.path.join(l2.daemon.lightning_dir, chainparams['name'], 'lightningd.sqlite3')
l2_db_path_bak = os.path.join(l2.daemon.lightning_dir, chainparams['name'], 'lightningd.sqlite3.bak')
copyfile(l2_db_path, l2_db_path_bak)
l2.start()
sync_blockheight(bitcoind, [l2])
# push some money from l3->l2, so that the commit counter advances
l2.rpc.connect(l3.info['id'], 'localhost', l3.port)
> l2.daemon.wait_for_log('now ACTIVE')
tests/test_closing.py:908:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
contrib/pyln-testing/pyln/testing/utils.py:288: in wait_for_log
return self.wait_for_logs([regex], timeout)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = <pyln.testing.utils.LightningD object at 0x7f0c145a32d0>
regexs = ['now ACTIVE'], timeout = 60
def wait_for_logs(self, regexs, timeout=TIMEOUT):
"""Look for `regexs` in the logs.
We tail the stdout of the process and look for each regex in `regexs`,
starting from last of the previous waited-for log entries (if any). We
fail if the timeout is exceeded or if the underlying process
exits before all the `regexs` were found.
If timeout is None, no time-out is applied.
"""
logging.debug("Waiting for {} in the logs".format(regexs))
exs = [re.compile(r) for r in regexs]
start_time = time.time()
pos = self.logsearch_start
while True:
if timeout is not None and time.time() > start_time + timeout:
print("Time-out: can't find {} in logs".format(exs))
for r in exs:
if self.is_in_log(r):
print("({} was previously in logs!)".format(r))
> raise TimeoutError('Unable to find "{}" in logs.'.format(exs))
E TimeoutError: Unable to find "[re.compile('now ACTIVE')]" in logs.
contrib/pyln-testing/pyln/testing/utils.py:264: TimeoutError
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And when it's set, and we're SLOW_MACHINE, simply disable valgrind.
Since Travis (SLOW_MACHINE=1) only does VALGRIND=1 DEVELOPER=1 tests,
and VALGRIND=0 DEVELOPER=0 tests, it was missing tests which needed
DEVELOPER and !VALGRIND.
Instead, this demotes them to non-valgrind tests for SLOW_MACHINEs.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I started replacing all get_node() calls, but got bored, so then just did the
tests which call get_node() 3 times or more.
Ends up not making a measurable speed difference, but it does make some
things neater and more standard.
Times with SLOW_MACHINE=1 (given that's how Travis tests):
Time before (non-valgrind):
393 sec (had 3 failures?)
Time after (non-valgrind):
410 sec
Time before (valgrind):
890 seconds (had 2 failures)
Time after (valgrind):
892 sec
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
