Closed skwp closed 2 years ago
I could reproduce the crash leading to the Internal Server Error.
TODO: Test with multiple nodes, disconnect the current one. UPDATE: Could not replicate the error here (tested with one Bitcoin and one Liquid node and with one Bitcoin main net and Bitcoin regtest node).
As for
Maybe warn or prevent Specter from registering competing nodes on the same port.
I checked, bitcoind already throws an error if you want to bind the RPC port twice (using different data dirs). But, a flash informing the user that he is saving a node with the same RPC port twice would definitely not hurt.
@skwp I took the liberty to rename the issue title, I couldn't reproduce the problem with multiple nodes only when having only one node.
Finally narrowed it down further: The internal server error was triggered, if a RPC connection went down and there were no wallets and then hitting "New wallet" button or refresh.
Describe the bug Specter loads with a white Internal Server Error screen on startup
To Reproduce
I believe this has to do with having multiple nodes registered that are all localhost. Trace is attached. There are a few problems here
Specter should capture all exceptions at the high level and show a graceful error screen. There should never be a white Internal server error screen shown.
Investigate the "can only be changed with one another" error log and figure out what's triggering it.
Figure out how Specter can fall back to or remember the active node if multiple nodes are registered and only one is active.
Maybe warn or prevent Specter from registering competing nodes on the same port. I believe something about that is causing this crash.