We are returning a `BADONION` error despite the cause being an invalid onion
payload containing an unknown even TLV type. It really should return
`INVALID_ONION_PAYLOAD` errors instead.
Add new check if we're funder trying to add HTLC, keeping us
with enough extra funds to pay for another HTLC the peer might add.
We also need to adjust the spendable_msat calculation, and update
various tests which try to unbalance channels. We eliminate
the now-redundant test_channel_drainage entirely.
Changelog-Fixed: Corner case where channel could become unusable (https://github.com/lightningnetwork/lightning-rfc/issues/728)
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Using it with a different value to the amount sent causes a crash in 0.8.0,
which is effectively deprecating it, so let's disallow it now.
Changelog-Changed: If the optional `msatoshi` param to sendpay for non-MPP is set, it must be the exact amount sent to the final recipient.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Detect a previously non-permanent error (`final_cltv_too_soon`) that has been merged into a permanent error (`incorrect_or_unknown_payment_details`), and retry that failure case in `pay`.
We still close the channel if we *send* an error, but we seem to have hit
another case where LND sends an error which seems transient, so this will
make a best-effort attempt to preserve our channel in that case.
Some test have to be modified, since they don't terminate as they did
previously :(
Changelog-Changed: quirks: We'll now reconnect and retry if we get an error on an established channel. This works around lnd sending error messages that may be non-fatal.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Thanks to @t-bast, who made this possible by interop testing with Eclair!
Changelog-Added: Protocol: can now send and receive TLV-style onion messages.
Changelog-Added: Protocol: can now send and receive BOLT11 payment_secrets.
Changelog-Added: Protocol: can now receive basic multi-part payments.
Changelog-Added: RPC: low-level commands sendpay and waitsendpay can now be used to manually send multi-part payments.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Bastien TEINTURIER <bastien@acinq.fr> writes:
> It looks like the split on c-lightning side is quite limited at the moment:
> the only option is to split a payment in exactly its two halves,
> otherwise I get rejected because of the rule of overpaying more than
> twice the amount?
We only tested exactly two equal-size payments; indeed, our finalhop
test was backwards. We only complain if the final hop pays more than
twice msat (technically, this test is still too loose for mpp: the
spec says we should sum to the exact amount).
Reported-by: @t-bast
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This won't usually be visible to the end-user, since the pay plugin doesn't
do multi-part yet (and mpp requires EXPERIMENTAL_FEATURES), but we're ready
once it does.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The invoice_try_pay code now takes a set, rather than a single htlc, but
it's basically the same thing.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Explicit #if EXPERIMENTAL_FEATURES check in case we enable them at different
times, but it requires a payment_secret since we put them in the same field.
This incidently stops it working on legacy nodes.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
msatoshi was used to indicate the amount the invoice asked for, but
for parallel sendpay it's required, as it allows our sanity check of
limiting the total payments in flight, ie. it becomes
'total_msat'.
There's a special case for sendonion, which always tells us the value is 0.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We currently refuse a payment if one is already in flight. For parallel
payments, it's a bit more subtle: we want to refuse if it we already have
the total-amount-of-invoice in flight.
So we get all the current payments, and sum the pending ones.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Now "raw_payload" is always the complete string (including realm or length
bytes at the front).
This has several effects:
1. We can receive an decrypt an onion which is grossly malformed.
2. We can still hand this to the htlc_accepted hook.
3. We then fail it unless the htlc_accepted accepts it manually.
4. The createonion API now takes the raw payload, and does not know
anything about "style".
The only caveat is that the sphinx code needs to know the payload
length: we have a call for that, which simply tells it to copy the
entire onion (and treat us as the final node) if it's invalid.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
If we can't decode the onion, because the onion got corrupted or we used
`sendonion` without specifying the `shared_secrets` used, the best we can do
is tell the caller instead.
This means that c-lightning can now internally decrypt an eventual error
message, and not force the caller to implement the decryption. The main
difficulty was that we now have a new state (channels and nodes not specified,
while shared_secrets are specified) which needed to be handled.
We are breaking with a couple of assumptions, namely that we have the
`path_secrets` to decode the error onion. If this happens we just want it to
error out.
We don't set the secret to compulsory (yet!) but put code in for the
future. Meanwhile, if there is a secret, check it is correct.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
1. Printed form is always "[<nodeid>-]<prefix>: <string>"
2. "jcon fd %i" becomes "jsonrpc #%i".
3. "jsonrpc" log is only used once, and is removed.
4. "database" log prefix is use for db accesses.
5. "lightningd(%i)" becomes simply "lightningd" without the pid.
6. The "lightningd_" prefix is stripped from subd log prefixes, and pid removed.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-changed: Logging: formatting made uniform: [NODEID-]SUBSYSTEM: MESSAGE
Changelog-removed: `lightning_` prefixes removed from subdaemon names, including in listpeers `owner` field.
Reduce test_feerate_stress iterations, and simply don't run
test_pay_retry under VALGRIND with SLOW_MACHINE at all.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Whenever we have multi-connected nodes, out-of-order gossip is possible.
In particular, if a node_announcement is 1 second fresher than the
channel_announcement, a timestamp_filter might get one and not the
other.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It's generally clearer to have simple hardcoded numbers with an
#if DEVELOPER around it, than apparent variables which aren't, really.
Interestingly, our pruning test was always kinda broken: we have to pass
two cycles, since l2 will refresh the channel once to avoid pruning.
Do the more obvious thing, and cut the network in half and check that
l1 and l3 time out.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Make update_local_channel use a timer if it's too soon to make another
update.
1. Implement cupdate_different() which compares two updates.
2. make update_local_channel() take a single arg for timer usage.
3. Set timestamp of non-disable update back 5 minutes, so we can
always generate a disable update if we need to.
4. Make update_local_channel() itself do the "unchanged update" suppression.
gossipd: clean up local channel updates.
5. Keep pointer to the current timer so we override any old updates with
a new one, to avoid a race.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're about to change the API, so this makes the tests still work
across the transition (and, as a bonus, tests our backwards compat
shim).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Remove gratuitous prints, add explanations of what's going on,
and demonstrate that we can add a final trimmed HTLC but not
a non-trimmed one.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Subtracting both arbitrarily reduces our capacity, even for ourselves
since the routing logic uses this maximum.
I also changed 'advertise' to 'advertize', since we use american
spelling.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Turns out we needed more comprehensive testing; we ended up with three
separate tests. To avoid changing test_channel_drainage as we fix
spendable_msat, I substituted raw numbers there.
The first is a variation of the existing tests, testing we can't
exceed spendable_msat, and we can pay it, both ways.
The second is with a larger amount, which triggers a different problem.
The final is with a giant channel, which tests our 2^32-1 msat cap.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is where payment tests should go. Also mark it xfail for the moment,
and remove developer-only tag (propagating gossip is only 60 seconds, which
is OK).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Broken by 909913c265, but since Travis
skips this test ("temporarily", according to the commit msg in January!)
it wasn't caught.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Here I add the test for this 5 local_failed case in this commit.
There 5 cases for FORWARD_LOCAL_FAILED status:
1. When Msater resolves the reply about the next peer infor(sent by Gossipd), and need handle unknown next peer failure in channel_resolve_reply();
2. When Master handle the forward process with the htlc_in and the id of next hop, it tries to drive a new htlc_out but fails in forward_htlc();
3. When we send htlc_out, Master asks Channeld to add a new htlc into the outgoing channel but Channeld fails. Master need handle and store this failure in rcvd_htlc_reply();
4. When Channeld receives a new revoke message, if the state of corresponding htlc is RCVD_ADD_ACK_REVOCATION, Master will tries to resolve onionpacket and handle the failure before resolving the next hop in peer_got_revoke();
5. When Onchaind finds the htlc time out or missing htlc, Master need handle these failure as FORWARD_LOCAL_FAILED in if it's forward payment case.
We generated blocks to announce the channel, but it can also expire
the HTLC if the timing is wrong. We don't need to anyway, since we
fixed the FIXME; we store local unannounced channels for restoration
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We're going to make it async, so start by moving the core code into
invoice.c and having that directly call fail/success functions for the
htlc.
We add an extra check in fulfill_htlc() that the HTLC state is correct:
that can't happen now, but may once we're async.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is currently done higher up, in handle_channel_update(), but
that's one reason why handle_channel_update() has to do a channel
lookup. Moving the check down means handle_channel_update() can do a
minimal "get node id for this channel" so it can check the signature.
This helps, because the chan lookup semantics are changing in the next
few patches.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
gossipd in l1 might not have registered l2 reconnecting, thus considering
the channel local-disabled.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
New name is less confusing, and most people should be transitioning to
listpays rather than this anyway.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the same deprecation, but one level up. For the moment, we
still support invoices with a `h` field (where description will be
necessary) but that will be removed once this option is removed.
Note that I just changed pylightning without backwards compatibility,
since the field was unlikely to be used, but we could do something
more complex here?
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is particularly interesting because we handle overflow during route
calculation now; this could happen in theory once we wumbo.
It fixes a thinko when we print out routehints, too: we want to print
them out literally, not print out the effect they have on fees (which
is in the route, which we also print).
This ABI change doesn't need a CHANGELOG, since paystatus is new since
release.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Little point having users handle the postfixes manually, this
translates them, and also allows Millisatoshi to be used wherever an
'int' would be previously.
There are also helpers to create the formatting in a way c-lightning's
JSONRPC will accept.
All standard arithmetic operations with integers work.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We had occasional failures, because the fuzz could overwhelm the difference
in routes. Increasing the amount to 2,000,000 millisatoshis makes the
riskfactor 53msat (2000000 * 14 * 10 / 5259600) which is always greater
than the worst-case fuzz of 5% on the fee of 1002msat.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
I got a spurious failure because the final node gave a CLTV error and
so it decided to use a different channel. It should probably handle
this corner case better, but meanwhile make the test robust.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It was waiting for a remote channel, but not for all the interesting
channels we want to check. It can sometimes happen that further away
channels are added before closer ones are added, depending on
propagation path, flush timers and bitcoind poll timers. This now just
checks for all channels, which also reduces the ambiguity of whether
we selected a path solely because we were lacking alternatives.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
Travis timed out.
Waiting for three fundchannel commands depends on the bitcoind polling
interval (30 seconds), and then waiting for gossip propagation
requires two propagation intervals (120 seconds).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Up until now, riskfactor was useless due to implementation bugs, and
also the default setting is wrong (too low to have an effect on
reasonable payment scenarios).
Let's simplify the definition (by assuming that P(failure) of a node
is 1), to make it a simple percentage. I examined the current network
fees to see what would work, and under this definition, a default of
10 seems reasonable (equivalent to 1000 under the old definition).
It is *this* change which finally fixes our test case! The riskfactor
is now 40msat (1500000 * 14 * 10 / 5259600 = 39.9), comparable with
worst-case fuzz is 50msat (1001 * 0.05 = 50).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The test sometimes passes: our routing logic always chooses between
the shorter of two equal-cost routes (because we compare best with <
not <=).
By adding another hop, we add more noise, and by making the alternate
route fee 0 we provide the worst case.
But to be fair, we make the amount of the payment ~50c (15,000,000
msat), and increase our cltv-delay to 14 and fee-base 1000 to match
mainnet. The final patch shows the effect of this choice.
Otherwise our risk penalty is completely in the noise on
mainnet which has the vast majority of fees set at 1000msat + 1ppm.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This is the direct cause of the failure of the original
test_pay_direct test and it makes sense: invoice routehints may not be
necessary, so try without them *first* rather than last.
We didn't mention the use of routehints in CHANGELOG at all yet, so
do that now.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
It is suddenly timing out a lot and is breaking master, so we
temporarily disable it until it is fixed.
Signed-off-by: Christian Decker <decker.christian@gmail.com>
So add a new 'strategy' field. This makes it clearer what is going
on, currently one of:
* "Initial attempt"
* "Excluded channel <scid>"
* "Removed route hint"
* "Excluded expensive channel <scid>"
* "Excluded delaying channel <scid>"
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We sanitize the routes: firstly, we assume appending so eliminate the
first hop if the route points straight to us. Secondly, eliminate empty
hints. Thirdly, trim overlong hints.
Then we just use the first route hint.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Currently only used by gossipd for channel elimination.
Also print them in canonical form (/[01]), so tests need to be
changed.
Suggested-by: @cdecker
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Funder can't spend the fee it needs to pay for the commitment transaction:
we were not converting to millisatoshis, however!
This breaks our routeboost test, which no longer has sufficient funds
to make payment.
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>
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>
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>
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.
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.
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>
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>
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>
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>
