djg
djg
The extension was the batch payments branch, but I don't think that was related -- it doesn't touch anything related to subscriptions afaik. The server was on main.
`git log` in the build: ``` * 676d400a sentientwaffle: fix(extension): dont "overpay" via partial payment - (HEAD -> dj-payment-scheduler-post, origin/dj-payment-scheduler-post)(6 weeks ago) * a96d44fa Nicholas Dudfield: chore(coil-extension): log rate -...
You're right, the slowness is most likely due to the routes propagating. I'll try and think of a fix.
I don't understand what jitter backoff would buy us -- as I understand it, it's meant for cases involving resource contention.
Only admins can update the balance manually, presumably so that they can set up accounts.
I think we already do that automatically -- when formatting, we first write the WAL, then we write the superblock (trailers then header). We could probably improve error messages. But...
@brson Was this all resolved by https://github.com/tigerbeetle/tigerbeetle/pull/1431?
> I pondered that, but as long we are not recording a new transfer to void the pending one (which would be the evidence of when we did it), it...
StateMachine's timestamps and timeouts are both nanoseconds. If timeouts were set with millisecond precision instead (leaving timestamps as nanoseconds), then we could guarantee that timeouts never land in the middle...
Ah, I see; thank you for the explanation! That does indeed make sense. > but the state machine can only deal with a limited number of them during each operation,...