Always print connectivity report #3677
Conversation
This is printed every 10 minutes, I see no reason why it shouldn't be in all the logs, it would give us valuable information about what is going on with node connectivity when validators come-back to us to report issues. Signed-off-by: Alexandru Gheorghe <[email protected]>
|
This can get pretty big if there are many unconnected authorities. Are we interested in exactly these? Because otherwise the metrics are always available per protocol - |
Right, but then there is also some network issue which would justify a bit of verbosity every 10 minutes. If we assume that these logs would be 1Meg in the worst case, then it would still take us 7 days to fill a GB. Sounds sensible for today's hardware. |
Your estimations might be right, but my question is mostly if we do this specifically to know the unconnected authorities or we are interested mainly in the connectivity ratio. |
Those metrics are good to tell you something went wrong, but they will fail to tell which nodes are not connecting or some other stuff. The flow I'm actually trying to optimise here, is people coming to us observing high level stuff is not working and providing us the log files they have, but there is literally nothing in there to tell you what went wrong. So at minimum we need to understand how well connected that node was by default.
My thoughts as well given how rarely we print this, I don't thing there is any danger there. |
The unnconnected list proved useful here #3314 (comment), so yeah I think we want it. |
famous last words :D |
|
Always possible to make a runlength encoded bitfield and then 7-bit or hex encode that, if youre worried about size. |
* master: (65 commits) collator protocol changes for elastic scaling (validator side) (#3302) Contracts use polkavm workspace deps (#3715) Add elastic scaling support in ParaInherent BenchBuilder (#3690) Removes `as [disambiguation_path]` from `derive_impl` usage (#3652) fix(paseo-spec): New Paseo Bootnodes (#3674) Improve Penpal runtime + emulated tests (#3543) Staking ledger bonding fixes (#3639) DescribeAllTerminal for HashedDescription (#3349) Increase timeout for assertions (#3680) Add subsystems regression tests to CI (#3527) Always print connectivity report (#3677) Revert "FRAME: Create `TransactionExtension` as a replacement for `SignedExtension` (#2280)" (#3665) authority-discovery: Add log for debugging DHT authority records (#3668) Construct Runtime v2 (#1378) Support for `keyring` in runtimes (#2044) Add api-name in `cannot query the runtime API version` warning (#3653) Add a PolkaVM-based executor (#3458) Adds default config for assets pallet (#3637) Bump handlebars from 4.3.7 to 5.1.0 (#3248) [Collator Selection] Fix weight refund for `set_candidacy_bond` (#3643) ...
This is printed every 10 minutes, I see no reason why it shouldn't be in all the logs, it would give us valuable information about what is going on with node connectivity when validators come-back to us to report issues. Signed-off-by: Alexandru Gheorghe <[email protected]>
|
Debug output is not to be printed using info logs. If people come to you complaining, you should either tell them which cli flags to change or come up for example with a dedicated RPC endpoint that generates debug information. This said, the change in this pr should be reverted. |
This is printed every 10 minutes, I see no reason why it shouldn't be in all the logs, it would give us valuable information about what is going on with node connectivity when validators come-back to us to report issues.