You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
https://opentracing.io gives a way for the lifetime of an RPC to be traced across multiple components in a distributed system. Think of this as the full solution to what debug transactions try to provide. It's also the one debugging tool that I miss having.
This will require RPCs to forward a UUID identifying the overall trace, and to forward that UUID along with any subsequent RPCs they make as a result.
Once we do this, it'd also be good to allow a UUID to be passed in from the client API for a transaction, so layers using opentracing could get the full view into FDB of what happens with their requests.
The text was updated successfully, but these errors were encountered:
The variant of this I used before (I think) buffered RPC traces in memory, which were periodically pushed to the trace collection service. I imagine Zipkin does something similar, but I haven't looked.
I'd be somewhat concerned if the answer was they get dumped to a log file, as that would put a significant limiter on how many traces we could propagate information for.
https://opentracing.io gives a way for the lifetime of an RPC to be traced across multiple components in a distributed system. Think of this as the full solution to what debug transactions try to provide. It's also the one debugging tool that I miss having.
This will require RPCs to forward a UUID identifying the overall trace, and to forward that UUID along with any subsequent RPCs they make as a result.
Once we do this, it'd also be good to allow a UUID to be passed in from the client API for a transaction, so layers using opentracing could get the full view into FDB of what happens with their requests.
The text was updated successfully, but these errors were encountered: