Do we have monitoring for when we get actual load?
@Anonymous + @Diego: add a graph to the dashboard that displays all the http status codes as returned by nginx
-
why? we need to be able to see if/when l1s are responding with error status codes, like 429s when any client(s), by IP address, hits the rate limit
-
like this graph
but the data is pulled from http response codes as reported by the l1s, not the http response codes as reported by arc widgets
@Diego: ask the ipfs gw team how they arrived at their rate limit settings, iirc 800reqs/s
@Diego: increase ssl_session_timeout value
Could we run a ML algorithm to improve the node weighting parameters to improve overall TTFB?
CPU usage is going to rise when there are more cache misses as we are doing inc. ver.
What techniques can we use to prevent Saturn from being down on an auto-update?
- We cant guarantee 100% uptime with just DNS alone.
- Service worker falls back to separate DNS address to map to second closest L1.
List of all L1s in a list
Does the IPFS Gateway support range requests on