Avoids failing the test with the pip warning:
WARNING: Running pip as the 'root' user can result in broken permissions and conflicting behaviour with the system package manager. It is recommended to use a virtual environment instead: https://pip.pypa.io/warnings/venv
reported by: @ksedgwic
Changelog-None
Fixes a bug in installer registration where executable is evaluated
before entrypoints and other details are added.
***RECKLESS STDERR***
Traceback (most recent call last):
File lightning/tools/reckless, line 382, in <module>
INSTALLERS['nodejs'].add_entrypoint('{name}')
KeyError: 'nodejs'
Reported by @ksedgwic
Changelog-None
When enabling or disabling a plugin, the entrypoint is inferred
from the user provided name. A canonical name should be used, which
the installer entrypoint formats help to determine (this generally strips
the file extension if one is provided.)
Also adds a timeout when testing a plugin. Previously the behavior
of pyln-client was relied upon to exit if not communicating with
lightningd, however, this behavior is not universal.
Changlelog-Changed: reckless now installs node.js plugins
Also removes support for pip editable install using pyproject.toml
`pip install -e .` This was a fallback method when a requirements
file was not present, but was hacky and often failed anyway.
reckless: remove installation via pyproject.toml
This method relied on pip install in editable mode (hacky) and often
failed to complete anyhow. We should instead encourage a requirements
file to be created/used for user installation.
This was pointed out by Daywalker [1]: we are synchronously processing
events from `lightningd` which means that if processing one of the
hooks or requests was slow or delayed, we would not get notifications,
hooks, or RPC requests, massively impacting the flexbility.
This highlights the issue with a failing test (it times out), and in
the next commit we will isolate event processing into their own task,
so to free the event loop from having to wait for an eventual
response.
[1] https://community.corelightning.org/c/developers/hold-invoice-plugin#comment_wrapper_16754493
The push bit was convenient for connectd to send our own gossip
to peers upon connecting by naively traversing the gossip_store
and sending anything flagged `push`. This function is now
performed by gossipd leaving no use for the push bit.
Changelog-Changed: `gossipd`: gossip_store PUSH bit is no longer set.
This implements the proposal to simply use timestamp as "all", "none"
or "stream". There's also a rough spec draft which I will post soon.
This *also* removes the last place where we would sometimes sweep the
entire gossip_store looking for their given timestamps.
We could also get rid of the actual timestamp filtering logic in
gossip_store_next if we want to, as it's now basically unused.
Changelog-Changed: Protocol: Simplify gossip_timestamp_filter handling to "all", "none" or "recent" instead of exact timestamp.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This removes the sweep logic as soon as they connect. This should save
connectd a significant number of CPU cycles and make @whitslack finally
stop hitting me.
Changelog-Changed: `connectd` no longer sweeps gossip_store file when peer connects, saving CPU for large nodes.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This was previously the role of connectd, but it's actually more
efficient for us to do it: connectd has to sweep through the entire
gossip_store, but we have datastructures for this already.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We were in fact feeding l1 its own gossip, which it doesn't ratelimit (this was
a bit fuzzy before, but definitely is the case now!).
So make this node actually l3, so we test what we expected to test.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This test makes l2 save db, make a payment, then rollback.
*Sometimes* under CI (but not here) we don't shutdown fast enough
after the payment, so the moves.json from the coin_movements.py
records it. Sure enough, the result is the node ends up with
a -10000msat balance.
Validated by putting a time.sleep(5) between:
```
l2.rpc.pay(inv['bolt11'])
import time
time.sleep(5)
# stop both nodes, roll back l2's database
```
The answer, of course, is to save and rollback *both* the db and
moves.json file.
Here's the error:
```
def test_penalty_htlc_tx_timeout(node_factory, bitcoind, chainparams):
...
assert account_balance(l3, channel_id) == 0
> assert account_balance(l2, channel_id) == 0
tests/test_closing.py:1527:
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
tests/utils.py:184: in account_balance
m_sum -= Millisatoshi(m['debit_msat'])
contrib/pyln-client/pyln/client/lightning.py:197: in __sub__
return Millisatoshi(int(self) - int(other))
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
self = -10000msat, v = -10000
...
if self.millisatoshis < 0:
> raise ValueError("Millisatoshi must be >= 0")
E ValueError: Millisatoshi must be >= 0
contrib/pyln-client/pyln/client/lightning.py:82: ValueError
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We need to wait until we're sure bcli has handed results to lightningd:
```
> assert feerates['perkw']['mutual_close'] == 5000
E assert 6250 == 5000
tests/test_misc.py:1617: AssertionError
```
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
This finally happened on my test box; tests simply stopped. Turns out
we were spinning here, with waitpid returning -1.
Since this is during shutdown, that also explains why pytest under CI
would hang, since timeouts don't apply during test teardown!
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
There are several benefits of doing this:
- prevent fuzz target bit rot
- more test coverage in CI
- greater visibility of fuzz tests, encouraging contributions to the
seed corpus and tests themselves
Turns out this was accidentally changed for v0.10.1 in
d8e68893f5 where we made fee levels less
aggressive.
Oops. I guess we can fix the docs. And we now have "2blocks" if you
want it really fast!
Closes: #6129
Changelog-Fixed: JSON-RPC: `feerates` document correctly that urgent means 6 blocks (not 2), and give better feerate examples.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
mempool.info gives sat/vB, which is 1000 too low for us!
See-also: #6161 (complains setting feerate to 5 doesn't work)
See-also: #6129
Suggested-by: @lightingorb
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
You still need to actually make a rune when lightningd starts, as
commando (for safety) won't work unless you actually generate a rune
(that it knows of!).
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Added: hsmtool: `makerune` command to make a master rune for a node.
1. Rename get_hsm_secret to get_unencrypted_hsm_secret.
2. Create a common helper for fetching full file contents.
3. Create new routine to decrypt if necessary: get_hsm_secret().
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
We were debugging a number of issues related to the forwarding logic,
when using public scids on private channels, and we noticed that we
are very verbose everywhere, except where it counts, i.e., what
decisions are being taken. So we add a couple of debug logs, and a
final info one that tells the operator which resolution was chosen in
the end.
This was always the intent, but now we have to reconstruct from the
disparate fields.
This means `option_anchor_outputs` is now redundant.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: Protocol: we will upfront reject channel_open which asks for a zeroconf channel unless we are going to do a zerconf channel.
I tested this indeed breaks if we don't accept it, then implemented
the code to accept it.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Changelog-Fixed: protocol: We now correctly accept the `option_scid_alias` bit in `open_channel` `channel_type`.
Changelog-Deprecated: protocol: Not setting `option_scid_alias` in `option_channel` `channel_type` for unannounced channels.
