Both of these plugins will fail in interesting ways, and we should
still handle them correctly.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
It wasn't JSON formatted either so there was no nice pretty-printing
way. This jsonifies and pretty prints it.
Signed-off-by: Christian Decker <@cdecker>
We no longer use `dev-override-fees` to set the fees, rather we
instrument the `bitcoind` proxy to return the desired feerates.
Signed-off-by: Christian Decker <@cdecker>
pytest was an indirect dependency so far, making that one
explicit, and the timeout plugin should allow us to kill a stuck test
before travis kills it, and thus allow us to see where it got stuck.
Signed-off-by: Christian Decker <@cdecker>
Because gossip in this case takes up to a minute, this test took 10
minutes. The workaround is to do the waiting-for-gossip all at once.
Now it takes 362 seconds.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Unlike other daemons, closingd doesn't listen to the master, but runs
simply to its own beat. So instead of responding to the JSON dev_memleak
command, we always check for memory leaks, and make sure that the
python tests fail if they see MEMLEAK in the logs.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We now keep multiple commands for a json_connection, and an array of
json_streams.
When a command wants to write something, we allocate a new json_stream
at the end of the array.
We always output from the first available json_stream; once that
command has finished, we free that and move to the next. Once all are
done, we wake the reader.
This means we won't read a new command if output is still pending, but
as most commands don't start writing until they're ready to write
everything, we still get command parallelism.
In particular, you can now 'waitinvoice' and 'delinvoice' and it will
work even though the 'waitinvoice' blocks.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We need to keep the remaining buffer, and we need to try to parse it
before we read the next. I first tried keeping it in the object, but
its lifetime is that of the *socket*, which we actually reopen for
every command.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This was hanging sometimes in travis, but actually checking the result
of the commands makes it *always* hang. We remove the waitinvoice
which will not return.
ZmnSCPxj points out that this behavior, introduced in
ce0bd7abd3, is a regression: it would be
nice to be able to cancel a waitinvoice. But that fix is more complex,
and will have to be another PR.
This test will now hang, but it's OK: we're about to fix it!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
When developing in regtest or testnet it is really inconvenient to
have to fake traffic and generate blocks just to get estimatesmartfee
to return a valid estimate. This just sets the minfee if bitcoind
doesn't return a valid estimate.
Reported-by: Rene Pickhardt <@renepickhardt>
Signed-off-by: Christian Decker <@cdecker>
In one case we can reduce, in the others we eliminated if VALGRIND.
Here are the ten slowest tests on my laptop:
469.75s call tests/test_closing.py::test_closing_torture
243.61s call tests/test_closing.py::test_onchain_multihtlc_our_unilateral
222.73s call tests/test_closing.py::test_onchain_multihtlc_their_unilateral
217.80s call tests/test_closing.py::test_closing_different_fees
146.14s call tests/test_connection.py::test_dataloss_protection
138.93s call tests/test_connection.py::test_restart_many_payments
129.66s call tests/test_gossip.py::test_gossip_persistence
128.73s call tests/test_connection.py::test_no_fee_estimate
122.46s call tests/test_misc.py::test_htlc_send_timeout
118.79s call tests/test_closing.py::test_onchain_dust_out
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
generate was deprecated some time ago, so we added the generate_block()
helper. But many calls crept back in, and git master refuses it.
(test_blockchaintrack relied on the return value, so make generate_block
return the list of blocks).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
After Ubuntu 18.10 upgrade, lots of new flake8 warnings.
$ flake8 --version:
3.5.0 (mccabe: 0.6.1, pycodestyle: 2.4.0, pyflakes: 1.6.0) CPython 3.6.7rc1 on Linux
Note it seems that W503 warned about line breaks before binary
operators, and W504 complains about them after. I prefer W504, so
disable W503.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Occasional failure in test_fulfill_incoming_first where the channel
closed before the final message from dev_disonnect was read. Cause
was the peer writing a gossip msg and failing due to ECONNRESET, before
it read the final message.
(Managed to reproduce under strace -f, FTW).
This is really a symptom of the fact that line_graph's announce=True
didn't wait for node announcements. Let's do that.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
When we have multiple HTLCs with the same preimage and the same CLTV,
it doesn't matter what order we treat them (they're literally
identical). But when we offer HTLCs with the same preimage but
different CLTVs, the commitment tx outputs look identical, but the
HTLC txs are different: if we simply take the first HTLC which matches
(and that's not the right one), the HTLC signature we got from them
won't match. As we rely on the signature matching to detect the fee
paid, we get:
onchaind: STATUS_FAIL_INTERNAL_ERROR: grind_fee failed
So we alter match_htlc_output() to return an array of all matching
HTLC indices, which can have more than one entry for offered HTLCs.
If it's our commitment, we loop through until one of the HTLC
signatures matches. If it's their commitment, we choose the HTLC with
the largest CLTV: we're going to ignore it once that hits anyway, so
this is the most conservative approach. If it's a penalty, it doesn't
matter since we steal all HTLC outputs the same independent of CLTV.
For accepted HTLCs, the CLTV value is encoded in the witness script,
so this confusion isn't possible. We nonetheless assert that the
CLTVs all match in that case.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We set up HTLCs with the same preimage and both different and same
CLTVs in both directions, then make sure that onchaind is OK and that
the HTLCs are failed without causing downstream failure.
We do this for both our-unilateral and their-unilateral cases.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Create a second HTLC with a different CTLV but same preimage; onchaind
uses the wrong signature and fails to grind it.
Reported-by: molz (#c-lightning)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This was suggested by Pierre-Marie as the solution to the 'same HTLC,
different CLTV' signature mismatch.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
My test case is a mainnet gossip store with 22107 channels, and
time to do `lightning-cli listchannels`:
Before: `lightning-cli listchannels` DEVELOPER=0
real 0m1.303000-1.324000(1.3114+/-0.0091)s
After:
real 0m0.629000-0.695000(0.64985+/-0.019)s
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This also highlights the danger of searching the logs: that error
appeared previously in the logs, so we didn't notice that the actual
withdraw call gave a different error.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And use wallet_forward_status_in_db() everywhere in db code.
And clean up extra CHANGELOG.md entry (looks like rebase error?)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Adapts the `test_forward_stats` test to include checks for the
`forwarded_payments` table. Will add checks for the `listforwardings`
RPC call next.
Signed-off-by: Christian Decker <@cdecker>
When the wrong key is used, the remote end simply hangs up.
We used to get a random errno, which tends to be "Operation now in progress."
Now it's defined to be 0, detect and provide a better error.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Have c-lightning nodes send out the largest value for
`htlc_maximum_msat` that makes sense, ie the lesser of
the peer's max_inflight_htlc value or the total channel
capacity minus the total channel reserve.
We don't create unannouncable channels, but other implementations can.
Not only is it rude to expose these via invoices, it's probably not
useable anyway.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There is an issue with the way we retrieve the channel accounting info
that will result in always showing 0 for all stats. This tests for it
and the next commit will fix it.
During tests, this is half our log! And Travis truncates it if we get
a failure in test_restart_many_payments.
Interestingly, test_logging had a bug which relied on this spam :)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't save them to the database, so fix things up as we load them.
Next patch will actually save them into the db, and this will become
COMPAT code.
Also: call htlc_in_check() with NULL on db load, as otherwise it aborts
internally.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Usually, we only close an incoming HTLC once the outgoing HTLC is completely
resolved. However, we short-cut this in the FULFILL case: we have the
preimage, so might as well use it immediately (in fact, we wait for it to
be committed, but we don't need to in theory).
As a side-effect of this, our assumption that every outgoing HTLC has
a corresponding incoming HTLC is incorrect, and this test (xfail) tickles
that corner case.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We split json_invoice(), as it now needs to round-trip to the gossipd,
and uniqueness checks need to happen *after* gossipd replies to avoid
a race.
For every candidate channel gossipd gives us, we check that it's in
state NORMAL (not shutting down, not still waiting for lockin), that
it's connected, and that it has capacity. We then choose one with
probability weighted by excess capacity, so larger channels are more
likely.
As a side effect of this, we can tell if an invoice is unpayble (no
channels have sufficient incoming capacity) or difficuly (no *online*
channels have sufficient capacity), so we add those warnings.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're about to slow down the invoice rpc (esp. under valgrind), which
breaks the delicate timing of the autocleaninvoice test.
Change that so the autocleaner (and timestamp) starts after the invoices
are added.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Travis failures:
valgrind: m_scheduler/sema.c:104 (vgModuleLocal_sema_down): Assertion 'sema->owner_lwpid != lwpid' failed.
host stacktrace:
==1296== at 0x38083F48: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x38084064: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x380841F1: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x38135DAE: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x380D328D: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x3809A4AC: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x3809AE43: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
==1296== by 0x380988CF: ??? (in /usr/lib/valgrind/memcheck-amd64-linux)
sched status:
running_tid=0
Thread 1: status = VgTs_WaitSys (lwpid 1296)
==1296== at 0x5729730: __poll_nocancel (syscall-template.S:84)
==1296== by 0x4348DF: daemon_poll (daemon.c:78)
==1296== by 0x4169E7: io_poll_lightningd (lightningd.c:543)
==1296== by 0x471ECD: io_loop (poll.c:282)
==1296== by 0x416E06: main (lightningd.c:744)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We have a lot of infrastructure to delay local channel_updates to
avoid spamming on each peer reconnect; we had to keep tracking of
pending ones though, in case we needed the very latest for sending an
error when failing an HTLC.
Instead, it's far simpler to set the local_disabled flag on a channel
when we disconnect, but only send a disabling channel_update if we
actually fail an HTLC.
Note: handle_channel_update() TAKES update (due to tal_arr_dup), but we
didn't use that before. Now we do, add annotation.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We trade channel_update before channel_announce makes the channel
public, and currently forget them when we finally get the
channel_announce. We should instead apply them, and not rely on
retransmission (which we remove in the next patch!).
This earlier channel_update means test_gossip_jsonrpc triggers too
early, so have that wait for node_announcement.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The help command now adds command usage to its output by calling each
command handler in CMD_USAGE mode.
Instead of seeing, for example:
decodepay
Decode {bolt11}, using {description} if necessary
we see:
decodepay bolt11 [description]
Decode {bolt11}, using {description} if necessary
Signed-off-by: Mark Beckwith <wythe@intrig.com>
Incrementing version number means stores which were prior to the previous
commit will be removed, and refreshed. The simplest fix, if not the most
efficient.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There's no reason for the db to ever return non-NULL if it's spent. And there's
only one caller, for which that is definitely true.
Suggested-by: @cdecker
Fixes: #1934
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We extract the tx from the logs, and then we wait until that hits
the mempool. This is more reliable than 'sendrawtx' in the logs,
which might catch a previous sendrawtx; it's also more explicit
that we expect that tx exactly.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
BOLT 7's been updated to split the flags field in `channel_update`
into two: `channel_flags` and `message_flags`. This changeset does the
minimal necessary to get to building with the new flags.
We currently just ignore them. This is one reason the hsm (in some places)
explicitly calls log_broken so we get some idea.
This was the only subdaemon which had a NULL msgcb and msgname, so eliminate
those checks in subd.c.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Got a spurious failure in test_no_fee_estimate; we fired too soon from the logs (presumably
we raced in on the first response, but estimatesmartfee gets called 3 times).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is a simple reverse proxy that `bitcoin-cli` can talk to when invoked by
`lightningd`. It allows us to trace `bitcoin-cli` calls, and intercept calls to
mock the replies, better than the current bash-script based method.
It's an array: we were only saving the single element; if there was more than
one changed HTLC we'd get a bad signature!
The report in #1907 is probably caused by the other side re-requesting
something we considered already finalized; to avoid this particular error,
we should set the field to NULL if there's no last_sent_commit.
I'm increasingly of the opinion we want to just save all the update
packets to the db and blast them out, instead of doing this
second-guessing dance.
Fixes: #1907
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
These happen after we compact the store; every log I've seen of a
restart on a real node has a message about truncating the store,
because node_announcements predate channel_announcements.
I extracted one such case from testnet, and reduced it to test here.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1. Wait for a 'sendrawtransaction' *after* the dev-fail message; don't be
fooled by a previous one.
2. Turning on estimate fee sets fees exactly; just wait for it to be processed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's probably unnecessary to have this weird way of injecting results
now we have explicit feerate args.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And, reluctantly, default to bitcoind style.
"It's wrong to be right too soon."
Suggested-by: @cdecker
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We could refine this later (based on existing wallet, for example), but
this gives some estimate.
[ Rename onchain_estimates -> onchain_fee_estimates Suggested-by: @SimonVrouwe ]
[ Factor of 1000 fix Reported-by: @SimonVrouwe ]
Suggested-by: @molxyz
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't know what our peer is doing, but if we see those values, maybe
they did too, and for longer. And add the min/max acceptable values
into our JSON API.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is useful mainly in the case where bitcoind is not giving estimates,
but can also be used to bias results if you want.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
And no more filtering out messages, as we should no longer spam the
logs with them (the 'Connected json input' one was removed some time
ago).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The comment was wrong: the channel being locked in was triggering
the fee update and hence the disconnect. But that can actually
happen before fund_channel returns, as that waits for the gossipd
to see the channel active.
Best to do the fee update manually, so it's exactly what we want.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
If feerates change, L2 sends L3 a commit for that, which causes us to
fail the assert (which says we won't send a commitment_signed).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We use feerate in several places, and each one really should react
differently when it's not available (such as when bitcoind is still
catching up):
1. For general fee-enforcement, we use the broadest possible limits.
2. For closingd, we use it as our opening negotiation point: just use half
the last tx feerate.
3. For onchaind, we can use the last tx feerate as a guide for our own txs;
it might be too high, but at least we know it was sufficient to be mined.
4. For withdraw and fund_channel, we can simply refuse.
Fixes: #1836
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Manipulate fees via fake-bitcoin-cli. It's not quite the same, as
these are pre-smoothing, so we need a restart to override that where
we really need an exact change. Or we can wait until it reaches a
certain value in cases we don't care about exact amounts.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't respond to fee changes until we're locked in: make sure we catch
up at that point.
Note that we use NORMAL fees during opening, but IMMEDIATE after, so
this often sends a fee update. The tests which break, we set those
feerates to be equal.
This (sometimes) changes the behavior of test_permfail, as we now
get an immediate commit, so that is fixed too so we always wait for
that to complete.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is a noop if we're opening a new channel (channel_fees_can_change(channel)
is false until funding locked in), but important if we're restarting.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We don't update a channel's feerate on reestablishment: we insert a restart
in test_onchain_different_fees() (which we'll need soon anyway) to show it.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
When in this state, we send a canned error "Awaiting unilateral close".
We enter this both when we drop to chain, and when we're trying to get
them to drop to chain due to option_data_loss_protect.
As this state (unlike channel errors) is saved to the database, it means
we will *never* talk to a peer again in this state, so they can't
confuse us.
Since we set this state in channel_fail_permanent() (which is the only
place we call drop_to_chain for a unilateral close), we don't need to
save to the db: channel_set_state() does that for us.
This state change has a subtle effect: we return WIRE_UNKNOWN_NEXT_PEER
instead of WIRE_TEMPORARY_CHANNEL_FAILURE as soon as we get a failure
with a peer. To provoke a temporary failure in test_pay_disconnect we
take the node offline.
Reported-by: Christian Decker @cdecker
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This means we don't try to unilaterally close after a restart, *and*
we can tell onchaind to try to use the point to recover funds when the
peer unilaterally closes.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Firstly, if they claim to know a future value, we ask the HSM; if
they're right, we tell master what the per-commitment-secret it gave
us (we have no way to validate this, though) and it will not broadcast
a unilateral (knowing it will cause them to use a penalty tx!).
Otherwise, we check the results they sent were valid. The spec says
to do this (and close the channel if it's wrong!), because otherwise they
could continually lie and give us a bad per-commitment-secret when we
actually need it.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Tests were failing when in the same thread after a test which set
log_all_io=True, because SIGUSR1 seemed to be turning logging *off*.
This is due to Python using references not copies for assignment.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is required for the next test, which has to log messages from channeld
as soon as it starts (so might be too late if it sends SIGUSR1).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We ignore incoming for now, but this means we advertize the option and
we send the required fields.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
peer features are only kept for connected peers (as they can change),
but we didn't update them on reconnect. The main effect was that
after a restart we displayed the features as empty, even after
reconnect.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1. l1 update_fee -> l2
2. l1 commitment_signed -> l2 (using new feerate)
3. l1 <- revoke_and_ack l2
4. l1 <- commitment_signed l2 (using new feerate)
5. l1 -> revoke_and_ack l2
When we break the connection after #3, the reconnection causes #4 to
be retransmitted, but it turns out l1 wasn't telling the master to set
the local feerate until it received the commitment_signed, so on
reconnect it uses the old feerate, with predictable results (bad
signature).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Useful it we want to intercept bitcoin-cli first.
We move the getinfo() caching into start(), as that's when we can actually
use RPC.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're going to use it to override specific commands. It's non-valgrinded
already since we use '--trace-children-skip=*bitcoin-cli*' so the overhead
should be minimal.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We got an index error, because status had only one field (onchaind not
started yet).
> wait_for(lambda: only_one(p.rpc.listpeers(l1.info['id'])['peers'][0]['channels'])['status'][1] == 'ONCHAIN:Tracking mutual close transaction')
E IndexError: list index out of range1
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Normal wallet txs get reconfirmed as blocks come in, but ones which need
closeinfo are more fragile, so we do it manually using txwatch for them.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
In one case, the channel_update which we expected to activate the channel
from l2 was suppressed as redundant. This is certainly valid, so just
check the results.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Waiting for three node_announcments isn't always enough, since l2 can
publish two of them (an independent bug). Do the more Right Thing and
just wait for 30 seconds of no input...
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1. connect convenience variable for improved readabilty.
2. a comment explaining that timer is on channel, not HTLC.
3. use modern python style in test_htlc_send_timeout
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Now sending a ping makes sense: it should force the other end to send
a reply, unblocking the commitment process.
Note that rather than waiting for a reply, we're actually spinning on
a 100ms loop in this case. But it's simple and it works.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Currently, if we don't realize a TCP connection is down, we almost
certainly don't find out until *after* we're sent the
commitment_signed message, in which case we cannot fail the incoming
HTLC.
This test demonstrates that. Note the 30 second sleep: we should really
run Travis tests in parallel!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
If we have an address hint, we start with that, but we'll use
node_announcement information if required.
Note: we (ab)use the address hint when restoring from the database
or reconnecting, even if the connection was *incoming*. That meant
that the recipient of a connection would *never* manage to connect out.
We still don't take multiple addresses from the DNS seeds: I assume we
should, since there could be IPv4 and IPv6.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
connectd tells master about every disconnection, and master knows
whether it's important to reconnect. Just get the master to invoke a new
connect command if it considers the peer important!
The only twist is timeouts: we don't want to immediately reconnect if
we've failed to connect. To solve this, connectd passes a 'delaytime'
to the master when a connection fails, and the master passes it back
when it asks for a connection.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
In particular, all opening_read_peer_msg() callers need to know there
was an error (presumably, negotiating) so they can stop, but we should
not exit.
This lets us reenable the final disabled test.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We now simply maintain a pubkey set for connected peers (we only care
if there's a reconnect), not the entire peer structure.
lightningd no longer queries us for getpeers: it knows more than we do
already.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There's now a potential race: the source peer connect returns, but in
destination peer the master hasn't read the connect message from
connectd, so the peer isn't in listpeers yet.
(Previously the connection stayed in connectd, so there was no such
window).
This is an occasional issue in a few places.
Note that we take the opportunity to speed up test_disconnectpeer too
while we're there.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Prior to this, lightningd would hand uninteresting peers back to connectd,
which would then return it to lightningd if it sent a non-gossip msg,
or if lightningd asked it to release the peer.
Now connectd hands the peer to lightningd once we've done the init
handshake, which hands it off to openingd.
This is a deep structural change, so we do the minimum here and cleanup
in the following patches.
Lightningd:
1. Remove peer_nongossip handling from connect_control and peer_control.
2. Remove list of outstanding fundchannel command; it was only needed to
find the race between us asking connectd to release the peer and it
reconnecting.
3. We can no longer tell if the remote end has started trying to fund a
channel (until it has succeeded): it's very transitory anyway so not
worth fixing.
4. We now always have a struct peer, and allocate an uncommitted_channel
for it, though it may never be used if neither end funds a channel.
5. We start funding on messages for openingd: we can get a funder_reply
or a fundee, or an error in response to our request to fund a channel.
so we handle all of them.
6. A new peer_start_openingd() is called after connectd hands us a peer.
7. json_fund_channel just looks through local peers; there are none
hidden in connectd any more.
8. We sometimes start a new openingd just to send an error message.
Openingd:
1. We always have information we need to accept them funding a channel (in
the init message).
2. We have to listen for three fds: peer, gossip and master, so we opencode
the poll.
3. We have an explicit message to start trying to fund a channel.
4. We can be told to send a message in our init message.
Testing:
1. We don't handle some things gracefully yet, so two tests are disabled.
2. 'hand_back_peer .*: now local again' from connectd is no longer a message,
openingd says 'Handed peer, entering loop' once its managing it.
3. peer['state'] used to be set to 'GOSSIPING' (otherwise this field doesn't
exist; 'state' is now per-channel. It doesn't exist at all now.
4. Some tests now need to turn on IO logging in openingd, not connectd.
5. There's a gap between connecting on one node and having connectd on
the peer hand over the connection to openingd. Our tests sometimes
checked getpeers() on the peer, and didn't see anything, so line_graph
needed updating.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
In particular, I found lightning_openingd processes after running
tests. When we use the dev_disconnect blackhole '0' option, they
stick around until the dev_disconnect file is truncated (there is only
so much you can do with only a file descriptor), so let's do that.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The following changes revealed this race, where expecting listchannels()
to contain two channels immediately after fund_channel() was racy.
We also derive the short_channel_id first, so we can search logs for the
exact messages.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The next patches get better at reconecting, so if we use dev-allow-localhost
nodes can often find each other and reconnect before shutting down; only
use that option where we actually need it.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Saw this in Travis: technically we return from the dev_set_max_scids...
cmd after sending it to gossipd, but we should wait for it to log.
Adding an internal reply message for a dev command seems overkill.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1. If the IPv6 address was public, that changed the wireaddr and thus the ipv4 bind
would not be to a wildcard and would fail.
2. Binding two fds to the same port on both wildcard IPv4 and IPv6 succeeds; we only
fail when we try to listen, so allow error at this point.
For some reason this triggered on my digital ocean machine.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
openingd calculates our reserve based on the channel amount (even if
we're funding, to keep the calculation in one place), but it wasn't
reporting it back to the master daemon. We initialized it to 0 so that
valgrind wouldn't get upset, as it's part of a structure we send over
the wire.
Have openingd report back, and also initialize it to an impossible value
as extra assurance. And remove a stray (harmless but weird) semicolon.
Reported-by: Gálli Zoltán
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This adds one line with the onion and the channel_update we extract from
it. This in turn allows us to check that the channel_update in the onion is not
type prefixed, and that we patch it correctly before passing it to gossipd.
The easiest way to do this is to play with the 'wallet_tx' semantics
and have 'amount' have meaning even when 'all_funds' is set.
Note that we change the string 'Cannot afford funding transaction' to
'Cannot afford transaction' as this code is also used for withdrawls.
Inspired-by: molz on #c-lightning
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The logs in various Travis failures show that it takes 20 seconds just for
closingd to read the init message. As a result, the close times out (default
is 30 seconds).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We were *supposed* to be waiting for the next commitment tx so we
made sure the one we broadcast was old, *but* the 'revoke_and_ack'
we were waiting for could be matched by the completion of the previous
'revoke_and_ack'.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This needs to be done separately from the rest of the daemon since we can
otherwise not make sure that it happens before the DB is freed and we might
still need the DN, and be running in a DB transaction, for some destructors to
run.
That was the cause of the bad gossip order failures: gossipd thought our
channel was live, but the other end didn't receive message last time.
Now gossipd doesn't use fd to kill us (connectd tells master to do so), we
can implement read_peer_msg_nogossip().
Fixes: #1706
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This patch guts gossipd of all peer-related functionality, and hands
all the peer-related requests to channeld instead.
gossipd now gets the final announcable addresses in its init msg, since
it doesn't handle socket binding any more.
lightningd now actually starts connectd, and activates it. The init
messages for both gossipd and connectd still contain redundant fields
which need cleaning up.
There are shims to handle the fact that connectd's wire messages are
still (mostly) gossipd messages.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Gossipd combines the information if it knows it, but that's really the
job of 'listnodes'. More importantly, channeld won't have access to
this information.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I saw an error in test_gossip_weirdalias in Travis, where listnodes(nodeid)
returned *BOTH* nodes; it happened to fail because [0] was the wrong one, but
it would have passed if the order had been different.
This helper asserts that we really do only have one element, and should
catch such bugs faster.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I could not figure out why test_announce_address suddenly stopped working:
I had previously been using DEVELOPER=1 on the cmdline for historical
reasons when testing locally.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
