[WIP] command: bitswap ledger #2814
Conversation
|
@whyrusleeping Would like your input if you think this is the right direction. I don't know if exporting |
|
@pfista the problem with working on this command now is that the ledger information isnt really used. Although i guess it might be interesting just so we can see how much data we've sent to each peer. I think instead of directly exporting the ledger, we should add a way to get a snapshot of a ledger. That way we avoid any risk of race conditions and keep things more well contained. |
|
This looks a lot cleaner. I'll need to take a better look in the morning, but LGTM right now 👍 |
|
It's worth mentioning too if this is not really needed anymore we should
|
|
@pfista i like this command. I just need to get to reviewing it (sorry about the delay, the decentralized web conference and nodeconf took a fair bit of time away from github) |
|
Subscribing. |
|
|
||
| func (e *Engine) LedgerSnapshot(p peer.ID) *LedgerSnapshot { | ||
| snapshot := &LedgerSnapshot{ | ||
| DebtRatio: e.findOrCreate(p).Accounting.Value(), |
There was a problem hiding this comment.
I think we're going to have to worry about locking a bit here. Lets make a single call to findOrCreate which is threadsafe on its own, then call lock on the returned ledger, defer the unlock, and then take the information we need from it
|
|
||
| func (e *Engine) LedgerSnapshot(p peer.ID) *LedgerSnapshot { | ||
| e.lock.Lock() | ||
| defer e.lock.Unlock() |
There was a problem hiding this comment.
@whyrusleeping Looking at the other code I noticed you're locking on the engine. Let me know if instead I should actually lock on the ledger like you mentioned in your previous comment.
There was a problem hiding this comment.
This will deadlock. findOrCreate takes the same lock that youre taking here. The correct way to do it will be to take the lock on the ledger you get back.
There was a problem hiding this comment.
Oh duh. I hadn't rebased yet, the old findOrCreate wasn't locking on engine...
License: MIT Signed-off-by: Michael Pfister <pfista@gmail.com>
License: MIT Signed-off-by: Michael Pfister <pfista@gmail.com>
License: MIT Signed-off-by: Michael Pfister <pfista@gmail.com>
pfista
force-pushed
the
437-bitswap-ledger
branch
from
2ff35ab
to
db1f6f0
Compare
|
closing in favor of #2852 |
Starting work for #437