Add ChannelDataPersister trait and point ChainMonitor to it. #681

valentinewallace · 2020-09-02T19:01:45Z

Intended to be a simple way for users to know where and how to put their
backup and persistence logic.

Sanity check/architecture reviews appreciated.
TODOS:

tests
I C&P'd the LinuxDataPersister straight from RLBRPC, prob should clean it up a bit
consider rebasing on Refactor chain monitoring #649
make it filesystem-agnostic using std::path
add fsync'ing the parent directory on write

Depends on #713 and #649.

codecov · 2020-09-02T19:31:27Z

Codecov Report

Merging #681 into master will increase coverage by 0.01%.
The diff coverage is 95.58%.

@@            Coverage Diff             @@
##           master     #681      +/-   ##
==========================================
+ Coverage   91.34%   91.36%   +0.01%     
==========================================
  Files          37       37              
  Lines       21974    22067      +93     
==========================================
+ Hits        20073    20162      +89     
- Misses       1901     1905       +4

Impacted Files	Coverage Δ
lightning/src/lib.rs	`100.00% <ø> (ø)`
lightning/src/ln/channelmanager.rs	`85.27% <ø> (ø)`
lightning/src/chain/chainmonitor.rs	`92.00% <75.00%> (-5.11%)`	⬇️
lightning/src/util/test_utils.rs	`88.63% <90.32%> (+0.06%)`	⬆️
lightning/src/chain/channelmonitor.rs	`95.67% <100.00%> (+0.15%)`	⬆️
lightning/src/ln/chanmon_update_fail_tests.rs	`97.56% <100.00%> (+0.10%)`	⬆️
lightning/src/ln/functional_test_utils.rs	`95.11% <100.00%> (+0.04%)`	⬆️
lightning/src/ln/functional_tests.rs	`96.97% <100.00%> (+<0.01%)`	⬆️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8a79877...cda8fb7. Read the comment docs.

devrandom · 2020-09-02T19:37:57Z

just confirming that there's no plan to persist anything before the funding tx is constructed. i.e. if the node restarts after a channel open but before funding_created, the channel is just forgotten.

valentinewallace · 2020-09-02T20:37:22Z

just confirming that there's no plan to persist anything before the funding tx is constructed. i.e. if the node restarts after a channel open but before funding_created, the channel is just forgotten.

Channel data is persisted at the same time as before this PR. Looking in ChannelManager, looks like we first persist on receipt of a FundingCreated or a FundingSigned message, which seems OK iiuc.

ariard

Approach ACK.

I still need to think about the consistency requirements with regards to higher layer of channel monitors but overall the flexibility offered is super cool :)

lightning-data-persister/src/lib.rs

lightning/src/ln/data_persister.rs

TheBlueMatt · 2020-09-09T19:46:30Z

lightning/src/chain/channelmonitor.rs

+		if let Some(orig_monitor) = monitors.get_mut(&outpoint) {
+			log_trace!(self.logger, "Updating Channel Monitor for channel {}", log_funding_info!(orig_monitor));
+			if let Err(e) = orig_monitor.update_monitor(&update, &self.broadcaster, &self.logger) {
+				return Err(e)


Per changes at https://github.com/rust-bitcoin/rust-lightning/pull/679/files#diff-dfe4c687889c83110048d0bb7a250ee2R147 we shouldn't return early, instead opting to store the new updated ChannelMonitor. Somewhat awkwardly, we probably need to only do so on some potential MonitorUpdateErrors - We should probably migrate towards those errors containing a flag to indicate if the situation implies a bogus update or simply a locked channel.e

Ok cool, I interpret this as: it'd be a good idea to add that "bogus update or locked channel" flag you mention in this PR.

lightning/src/chain/channelmonitor.rs

arik-so · 2020-09-11T09:35:35Z

Looking great so far to me, bu definitely curious about the persistence pipeline for memory -> disk during state negotiation procedures. I think that is where revisiting incremental updates would be very helpful, as those ought to be way less performance-heavy.

lightning/src/chain/channelmonitor.rs

lightning-data-persister/Cargo.toml

lightning-data-persister/src/lib.rs

joemphilips · 2020-09-13T11:41:28Z

This is great. My progress had been blocked with the lack of this feature in RL.
Looking forward to this change. (I'm not sure I can do meaningful review by my self, though)

Previously I could not know precisely when I should take SCB, because when a channel is closed there is no way to know about it besides polling the ChannelManager.
Hopefully this PR will ease my case.

valentinewallace · 2020-09-15T02:27:48Z

This is great. My progress had been blocked with the lack of this feature in RL.
Looking forward to this change. (I'm not sure I can do meaningful review by my self, though)

Previously I could not know precisely when I should take SCB, because when a channel is closed there is no way to know about it besides polling the ChannelManager.
Hopefully this PR will ease my case.

Hmm this PR isn't doesn't have the feature of notifying you on channel close (well, technically you'll get a callback on force close, but not on coop close). I filed #695

jkczyz

Running out of steam tonight. Will continue reviewing tomorrow.

lightning/src/ln/data_persister.rs

lightning/Cargo.toml

fuzz/src/chanmon_consistency.rs

jkczyz · 2020-09-22T01:24:39Z

lightning/src/ln/data_persister.rs

+/// supposed to be persisting bytes of data, backup sources may be considered
+/// "in consensus" if a majority of them agree on the current state and are on
+/// the highest known commitment number. See individual methods for more info.
+pub trait ChannelDataPersister: Send + Sync {


At the risk of being pedantic: how about simply ChannelPersister? In other words, would removing the word "data" lead to any ambiguity? Or should this rather be called ChannelMonitorPersister if it needs to differentiate from persisting Channels?

From a different angle, data_persister as used here as a module name and later as a variable name is ambiguous when reading code. Generally speaking, "data" usually can be substituted with some abstraction (e.g. channel) to provide clarity.

Hm, should we factor in that we intend to persist other data through this module in the future as well (e.g. ChannelManagers, Channels, etc)? Maybe it should just be called the Persister?

I would say that ChannelMonitor and Channel should be persisted atomically together, otherwise inconsistency could result. So that could be part of this trait.

Perhaps the non-channel-specific things in ChannelManager should be persisted using a different trait. That could be named e.g. NodePersister or such, since it includes non-channel relevant things like the announcement serial number.

That's a good point. The thing is that iiuc, stuff that needs to be persisted is pretty much divided into ChannelMonitors (which need to be time sensitively persisted whenever a channel is updated) and everything else, which should be persisted sometimes and would suck to lose but isn't at the same level of sensitivity.

How often the user wants to persist the less-sensitive stuff is subjective. We could make that decision for them (like the ChannelDataPersister does) but different users may have different fault-tolerance preferences. Not sure what the solution is here.

Ok discussed offline, update:

There's only 2 structs that need to be persisted: (1) the ChannelMonitors, and (2) the ChannelManager. The ChannelMonitors are sensitive data, and them being out of date could result in funds lost. However, the ChannelManager is not sensitive, and if e.g. a node starts up with an out of date ChannelManager, this will result in channels being force closed, but there's no chance for loss of funds.

So basically this PR is gonna focus on persisting ChannelMonitors, and we'll figure out how ChannelManager persistence is gonna work later. And the remaining open question is what to name the ChannelDataPersister.

I see several fields in Channel that might be critical for correct functioning so need persistence, including:

channel_state

channel_outbound

value_to_self_msat

short_channel_id (also needed to restore ChannelHolder correctly)

and maybe others. They probably won't cause loss of funds if out of date, but the system just won't work when restarted and restored from disk.

Also, ChannelHolder might have information that is required in order to be able to claim HTLCs, so if those are lost we can lose the HTLC value.

If we lose ChannelManager.last_node_announcement_serial, what is the impact?

Lastly, the KeysManager inside the ChannelManager has some state, including channel_child_index and rand_bytes_child_index. I'm not sure what is the impact if we end up with the same keyset for two channels.

and maybe others. They probably won't cause loss of funds if out of date, but the system just won't work when restarted and restored from disk.

You need to store the whole channel policy as announced to your counterparty, including its own policy, otherwise you may have channel policy discrepancies and your counterparty closing the channel unexpectedly (e.g a max_accepted_htlcs violation).

Also, ChannelHolder might have information that is required in order to be able to claim HTLCs, so if those are lost we can lose the HTLC value.

Good point, you need to store HTLC-relay data too, otherwise you may lost an in-flight update_fail_htlc form an downstream channel to an upstream one.

I don't think we need to store the Events as they should be re-generated if we re-apply the state correctly to every component.

If we lose ChannelManager.last_node_announcement_serial, what is the impact?

BOLT7:

if node_id is NOT previously known from a channel_announcement message, OR if timestamp is NOT greater than the last-received node_announcement from this node_id: SHOULD ignore the message.

Lastly, the KeysManager inside the ChannelManager has some state, including channel_child_index and rand_bytes_child_index. I'm not sure what is the impact if we end up with the same keyset for two channels.

Fund safety, so we should be careful storing it.

jkczyz · 2020-09-22T01:25:44Z

lightning/src/ln/data_persister.rs

+	///
+	/// [`update_id`]: struct.ChannelMonitorUpdate.html#structfield.update_id
+	/// [`ChannelMonitor::write_for_disk`]: ../../chain/channelmonitor/struct.ChannelMonitor.html#method.write_for_disk
+	fn persist_channel_data(&self, id: OutPoint, data: &ChannelMonitor<Self::Keys>) -> Result<(), ChannelMonitorUpdateErr>;


Depending on the outcome of the above naming discussion, consider naming this persist_new_channel.

In our discussion in the call, it seemed like our current idea of what our sample module will look like is: on persist_channel_data, the monitor update will be somehow persisted, and persistence of the whole ChannelManager will be kicked off in the background.

Given this, maybe Persister is the best name, and then indeed we can rename these methods persist_new_channel / update_channel?

IIUC, ChainMonitor will use the persister as currently implemented in this PR. ChannelManager can fire an event for persisting the ChannelManager that users can then process in the background. The latter can be implemented in a follow-up PR.

Regarding naming, Persister is probably fine so long as we don't need another trait to persist something else. Or possibly simply Persist to follow the loose rust trait naming conventions, which I went with in the chain module traits. Additionally, this trait could be moved to channelmonitor.rs which would make for a less ambiguous, module-qualified name: channelmonitor::Persist.

I don't have strong opinion yet on the function names. IIRC, we even discussed renaming ChannelMonitor (which could affect these function names), but I'm forgetting where that conversation went. I think I was opposed to it but can't recall the alternative name. :)

I switched to the names you suggested and channelmonitor::Persist. I think it's much nicer.

lightning/src/chain/channelmonitor.rs

jkczyz · 2020-09-22T05:53:52Z

lightning/src/chain/channelmonitor.rs

+pub struct MonitorUpdateError {
+	/// The human-readable error message string.
+	pub msg: &'static str,
+	/// A monitor update may have an error but still need to be persisted to disk.
+	pub persist_updated_monitor: bool
+}


Could we make this error an enum instead of adding a bool field. Not sure about naming but maybe something akin to the variants in ChannelMonitorUpdateErr.

Lmk what you think of the naming.

Also, w/r/t the naming of ChannelMonitorUpdateErr and MonitorUpdateError -- what do you think of just renaming MonitorUpdateError to InternalMonitorErr?

Based on our offline discussion, I'm starting to become of the mind that these should be reduced to a single error type. My reasoning is as follows:

ChannelMonitor's methods can return errors.

chain::Watch's interface is defined in terms of ChannelMonitor, ChannelMonitorUpdate, and ChannelMonitorUpdateErr.

ChainMonitor is an implementation of chain::Watch, but there may be other implementations of the trait.

Therefore, whether a ChannelMonitor must be persisted isn't necessarily always an internal detail handled by ChainMonitor; another implementation of chain::Watch or someone using ChannelMonitor directly would also need to do so.

My thought then would be to have one error type where the PermanentFailure variant has a bool indicating if it should be persisted.

That said, I'd like to get a better idea about where in the code that the "failure but persist" error will occur. I couldn't quite make that out from #681 (comment). Additionally, the docs for PermanentFailure indicate it can be used in two different ways. This makes me wonder if it should then be two different variants (and if so would this would remove the need for the bool?).

Thoughts? Let me know if I misunderstood anything.

Hmm, I think they're definitely very different in their purpose, though I don't necessarily think that implies they can't be the same type -

MonitorUpdateError is an error type returned by RL to the user to indicate that the update provided to the monitor was somehow bogus. It instructs the user whether they should still persist the updated object to disk, but in any case the update was not fully applied and ready to move forward.

ChannelMonitorUpdateErr is an error type generated by the user instructing the Channel{,Manager} what the current state of the monitor update is. It is maybe more aptly named ChannelMonitorUpdateStatus (and replacing the Result with just a general done/in-flight/failed enum).

The should-persist flag is definitely very awkward, and we should take pains in the documentation to point out that it is only relevant if you are running in a multi-primary redundant ChannelMonitor setup where each ChannelMonitor can unilaterally broadcast a transaction. In other words, for most folks, we can pretend its not there. See #679 for the one case where its ever set/matters. For normal folks, the MonitorUpdateError is just a human-readable error string that's arguably only useful for opening an issue on our github.

Indeed, for more complicated setups, arguably, they start to blend a bit more, but I do think there's value in them being different.

That's a good observation. I like where you're going with making ChannelMonitorUpdateErr a status instead of an error.

May need to think on this more, but just to throw an idea out there: would it make sense to have the chain::Watch interface return something like Result<ChannelMonitorUpdateStatus, MonitorUpdateError>? Possibly renaming these in some way, of course.

Result<ChannelMonitorUpdateStatus, MonitorUpdateError>

Hmm, interesting point. I suppose we could, though it doesn't mean anything to the ChannelManager which receives it - Err(_) would be converted to a PermanentFailure status either way.

Result<ChannelMonitorUpdateStatus, MonitorUpdateError>

Hmm, interesting point. I suppose we could, though it doesn't mean anything to the ChannelManager which receives it - Err(_) would be converted to a PermanentFailure status either way.

Not 100% sure if I follow that last point. Presumably, ChannelManager would process either the status if the Result is ok or the error if it is err using handle_monitor_err!.

It's the ChainMonitor that currently converts any MonitorUpdateError to a PermanentFailure. In what I'm floating, ChainMonitor would examine the Result and persist if needed and then simply return the Result as is back to ChannelManager.

Right, my point was only that that would be communicating extra information to the ChannelManager that it doesnt care about - PermanentFailure is the same to it as Err(_).

lightning/src/chain/channelmonitor.rs

lightning-data-persister/src/lib.rs

TheBlueMatt

New code looks pretty good, a few comments and I still need to thoroughly scan the new documentation, but can do after 649. Don't have strong feelings on naming.

lightning-data-persister/Cargo.toml

lightning-data-persister/src/lib.rs

lightning/src/chain/channelmonitor.rs

TheBlueMatt · 2020-09-24T18:47:08Z

lightning/src/chain/channelmonitor.rs

+			}
+			if should_persist {
+				match self.data_persister.update_channel_data(outpoint, &update, orig_monitor) {
+					Err(e) => return Err(e),


This must not "upgrade" a PermanentFailure in res to a TemporaryFailure.

lightning-data-persister/src/lib.rs

TheBlueMatt · 2020-09-26T00:41:37Z

lightning-data-persister/src/lib.rs

+		{
+			let mut f = fs::File::create(&tmp_filename)?;
+			monitor.write_for_disk(&mut f)?;
+			f.sync_all()?;


For this to be sensible on OSX, it looks like you need at least rust stdlib 1.36, at least going by this commit: rust-lang/rust@d602a6b. This should be documented somwehre.

Hm, I documented this but only by leaving a comment. Do you think it'd be ideal to add a README for gotchas like this? (It'd be somewhat redundant to the comments in the file.)

ariard

I think this really needs to be fixed, otherwise remove_dir_all is wiping out all the intent of persister as it doesn't check for emptiness ?

New tests sound good.

lightning/src/ln/chanmon_update_fail_tests.rs

lightning-persister/src/lib.rs

valentinewallace · 2020-10-13T17:17:19Z

I think this really needs to be fixed, otherwise remove_dir_all is wiping out all the intent of persister as it doesn't check for emptiness ?

Yeah really good catch that Jeff also noticed, fixed.

ariard

Tested ACK ed6101f

lightning-persister/src/lib.rs

TheBlueMatt

Looks good to me. A few nits, only really worth addressing if you're squashing/making other changes anyway. What say you @jkczyz

lightning/src/chain/channelmonitor.rs

fuzz/src/chanmon_consistency.rs

lightning/src/chain/chainmonitor.rs

The utilities will still be part of cfg(test). The purpose of this is to enable other crates in the workspace to use the test utilities.

Implementors of Persist are responsible for persisting new ChannelMonitors and ChannelMonitor updates to disk/backups.

- The ChainMonitor should: Whenever a new channel is added or updated, these updates should be conveyed to the persister and persisted to disk. Even if the update errors while it's being applied, the updated monitor still needs to be persisted.

jkczyz

Looks good to me. A few nits, only really worth addressing if you're squashing/making other changes anyway. What say you @jkczyz

LGTM. Sorry about the delay! Feel free to address any comments in a follow-up, if needed.

jkczyz · 2020-10-16T16:31:01Z

lightning-persister/src/lib.rs

+	// Test failure to rename in the process of atomically creating a channel
+	// monitor's file. We induce this failure by making the `tmp` file a
+	// directory.
+	// Explanation: given "from" = the file being renamed, "to" = the
+	// renamee that already exists: Windows should fail because it'll fail
+	// whenever "to" is a directory, and Unix should fail because if "from" is a
+	// file, then "to" is also required to be a file.


Was just thinking for either test, the presence of a read-only file ("to" or "from" depending on the test) might be sufficient to trigger the case without using a directory. Using a directory is an extra detail for the reader (different type vs invalid permission), so if possible it may be better to avoid.

jkczyz · 2020-10-16T16:40:33Z

lightning-persister/src/lib.rs

+		// Set up a dummy channel and force close. This will produce a monitor
+		// that we can then use to test persistence.
+		let chanmon_cfgs = create_chanmon_cfgs(2);
+		let node_cfgs = create_node_cfgs(2, &chanmon_cfgs);
+		let node_chanmgrs = create_node_chanmgrs(2, &node_cfgs, &[None, None]);
+		let nodes = create_network(2, &node_cfgs, &node_chanmgrs);
+		let chan = create_announced_chan_between_nodes(&nodes, 0, 1, InitFeatures::known(), InitFeatures::known());
+		nodes[1].node.force_close_channel(&chan.2);
+		let mut added_monitors = nodes[1].chain_monitor.added_monitors.lock().unwrap();


To avoid this setup, could you test the persister using write_channel_datalike the earlier tests? Likewise in the next test.

Hmm, it's nice to be able to test that it returns a PermanentError (since ultimately that's what we care about)

lightning-persister/src/lib.rs

Intended to be a cross-platform implementation of the channelmonitor::Persist trait. This adds a new lightning-persister crate, that uses the newly exposed lightning crate's test utilities. Notably, this crate is pretty small right now. However, due to future plans to add more data persistence (e.g. persisting the ChannelManager, etc) and a desire to avoid pulling in filesystem usage into the core lightning package, it is best for it to be separated out. Note: Windows necessitates the use of OpenOptions with the `write` permission enabled to `sync_all` on a newly opened channel's data file.

If a persister returns a temporary failure, the channel monitor should be able to be put on ice and then revived later. If a persister returns a permanent failure, the channel should be force closed.

This function does not necessarily write to disk, it can serialize to anything that implements Writer.

valentinewallace · 2020-10-16T19:09:41Z

One or two tests may need a small follow-up, but looks like mostly all comments addressed :)

…gdevkit#681) Previously we'd ignored generic arguments in traits, leading to bogus code generation after the Persister trait was added in lightningdevkit#681. This adds minimal support for it, fixing code generation on latest upstream.

…sister trait

…gdevkit#681) Previously we'd ignored generic arguments in traits, leading to bogus code generation after the Persister trait was added in lightningdevkit#681. This adds minimal support for it, fixing code generation on latest upstream.

…sister trait

Bindings Updates for #681 and more Clones

valentinewallace marked this pull request as draft September 2, 2020 19:01

jkczyz self-requested a review September 2, 2020 19:04

valentinewallace force-pushed the data-persister-refactor branch from 595db8a to 5bb24fb Compare September 3, 2020 20:14

ariard reviewed Sep 7, 2020

View reviewed changes

lightning-data-persister/src/lib.rs Outdated Show resolved Hide resolved

lightning/src/ln/data_persister.rs Outdated Show resolved Hide resolved

lightning/src/ln/data_persister.rs Outdated Show resolved Hide resolved

lightning/src/ln/data_persister.rs Outdated Show resolved Hide resolved

TheBlueMatt reviewed Sep 9, 2020

View reviewed changes

valentinewallace changed the title ~~Add ChannelDataPersister trait and point SMCM to it.~~ Add ChannelDataPersister trait and point ChainMonitor to it. Sep 10, 2020

julianknutsen reviewed Sep 13, 2020

View reviewed changes

valentinewallace force-pushed the data-persister-refactor branch from 5bb24fb to bd7e4b9 Compare September 15, 2020 00:55

joemphilips mentioned this pull request Sep 15, 2020

Notify users on channel close #695

Closed

jkczyz reviewed Sep 15, 2020

View reviewed changes

valentinewallace force-pushed the data-persister-refactor branch 2 times, most recently from 35b34be to 3ae7184 Compare September 16, 2020 22:13

valentinewallace marked this pull request as ready for review September 16, 2020 23:31

valentinewallace force-pushed the data-persister-refactor branch from ff43ee6 to 627871c Compare September 17, 2020 22:36

jkczyz reviewed Sep 22, 2020

View reviewed changes

devrandom reviewed Sep 22, 2020

View reviewed changes

lightning-data-persister/src/lib.rs Outdated Show resolved Hide resolved

valentinewallace force-pushed the data-persister-refactor branch from edb2855 to 4ff66ef Compare September 22, 2020 21:28

valentinewallace mentioned this pull request Sep 23, 2020

Refactor chain monitoring #649

Merged

TheBlueMatt reviewed Sep 24, 2020

View reviewed changes

devrandom reviewed Sep 25, 2020

View reviewed changes

lightning-data-persister/src/lib.rs Outdated Show resolved Hide resolved

lightning-data-persister/src/lib.rs Outdated Show resolved Hide resolved

TheBlueMatt reviewed Sep 26, 2020

View reviewed changes

TheBlueMatt added this to the 0.0.12 milestone Sep 27, 2020

valentinewallace force-pushed the data-persister-refactor branch 2 times, most recently from 5cc5671 to 2d88541 Compare September 30, 2020 02:06

ariard reviewed Oct 12, 2020

View reviewed changes

lightning/src/ln/chanmon_update_fail_tests.rs Outdated Show resolved Hide resolved

lightning-persister/src/lib.rs Show resolved Hide resolved

valentinewallace force-pushed the data-persister-refactor branch 2 times, most recently from fc5d755 to d546c91 Compare October 13, 2020 17:09

valentinewallace force-pushed the data-persister-refactor branch 2 times, most recently from 07de61f to ed6101f Compare October 13, 2020 19:29

ariard reviewed Oct 14, 2020

View reviewed changes

lightning-persister/src/lib.rs Outdated Show resolved Hide resolved

TheBlueMatt approved these changes Oct 15, 2020

View reviewed changes

lightning/src/chain/channelmonitor.rs Outdated Show resolved Hide resolved

fuzz/src/chanmon_consistency.rs Outdated Show resolved Hide resolved

lightning/src/chain/chainmonitor.rs Show resolved Hide resolved

valentinewallace added 3 commits October 16, 2020 11:30

Put test utilities behind a feature flag.

4350cc6

The utilities will still be part of cfg(test). The purpose of this is to enable other crates in the workspace to use the test utilities.

Add Persist trait.

ec4715c

Implementors of Persist are responsible for persisting new ChannelMonitors and ChannelMonitor updates to disk/backups.

Integrate Persist into ChainMonitor.

9c3f3e7

- The ChainMonitor should: Whenever a new channel is added or updated, these updates should be conveyed to the persister and persisted to disk. Even if the update errors while it's being applied, the updated monitor still needs to be persisted.

valentinewallace force-pushed the data-persister-refactor branch from d3aeaf8 to 8a9e1f3 Compare October 16, 2020 15:32

jkczyz approved these changes Oct 16, 2020

View reviewed changes

valentinewallace added 4 commits October 16, 2020 13:41

Test that Persist temp and perm failures behave as expected.

a8e82cb

If a persister returns a temporary failure, the channel monitor should be able to be put on ice and then revived later. If a persister returns a permanent failure, the channel should be force closed.

Rename ChannelMonitor::write_for_disk --> serialize_for_disk

fc68afb

This function does not necessarily write to disk, it can serialize to anything that implements Writer.

Fix clippy

cda8fb7

valentinewallace force-pushed the data-persister-refactor branch from 8a9e1f3 to cda8fb7 Compare October 16, 2020 17:41

TheBlueMatt merged commit 9c7c3b9 into lightningdevkit:master Oct 16, 2020

valentinewallace mentioned this pull request Oct 20, 2020

ChannelManager persistence #743

Closed

TheBlueMatt added a commit to TheBlueMatt/rust-lightning that referenced this pull request Nov 13, 2020

Update bindings demo for changes from lightningdevkit#681 and the Per…

f659a1c

…sister trait

TheBlueMatt mentioned this pull request Nov 13, 2020

Bindings Updates for #681 and more Clones #749

Merged

TheBlueMatt added a commit to TheBlueMatt/rust-lightning that referenced this pull request Nov 23, 2020

Update bindings demo for changes from lightningdevkit#681 and the Per…

90b668e

…sister trait

TheBlueMatt added a commit that referenced this pull request Nov 23, 2020

Merge pull request #749 from TheBlueMatt/2020-11-bindings-clones

52673d4

Bindings Updates for #681 and more Clones

jkczyz added this to In progress in Sample Implementations via automation Jan 27, 2021

jkczyz moved this from In progress to Done in Sample Implementations Jan 27, 2021

valentinewallace mentioned this pull request Mar 15, 2021

Sample ManyChannelMonitor #628

Closed

Add ChannelDataPersister trait and point ChainMonitor to it. #681

Add ChannelDataPersister trait and point ChainMonitor to it. #681

Conversation

valentinewallace commented Sep 2, 2020 • edited

codecov bot commented Sep 2, 2020 • edited

Codecov Report

devrandom commented Sep 2, 2020

valentinewallace commented Sep 2, 2020

ariard left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

arik-so commented Sep 11, 2020

joemphilips commented Sep 13, 2020

valentinewallace commented Sep 15, 2020

jkczyz left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

TheBlueMatt left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

ariard left a comment

Choose a reason for hiding this comment

valentinewallace commented Oct 13, 2020

ariard left a comment

Choose a reason for hiding this comment

TheBlueMatt left a comment

Choose a reason for hiding this comment

jkczyz left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

valentinewallace commented Oct 16, 2020

valentinewallace commented Sep 2, 2020 •

edited

codecov bot commented Sep 2, 2020 •

edited