Closed jarindr closed 5 years ago
I'd like to know more about configuration of the machine and was there Watcher instance running on the same machine or it was dedicated to the DB only.
@jarindr @jbunce - can we close this prehistoric issue? A lot has changed since Ari
, I guess we already have much better metrics, I can't recall other WatcherDB issues.
Feel free to close without further discussion 🤓 Thx
This incident happen in devcon4, while the blockexplorer polling the
/getTransactions
endpoint with 1 second interval on client, it overloaded the postgres instance and finally killed it cc. @jbunce .From the information we have, the amount of load shouldn't be enough to break the postgres / watcher, @jbunce even upgraded the instance but still the polling managed to push the CPU up to close to ~100%