Some issues remaining at the end of #128. However, the long validation process made it safer to push fixes as followups
Remove the *pImpl parameter from Retries() in conduit/pipeline/common.go
Per @winder
breaking out the context.Context out into a parameter, I was specifically looking for it to make sure the wait is properly terminated.
also breaking out RetryCount and logger. It's already a pure function in how it's used, but the signature doesn't guarantee that it's pure.
Alternatively, just make Retries private
Trim the special end of round message
In the final EC2 test, the size of the log at catchup grew from 14.1 GB to 16.2 GB. This is attributable to the change of the end of round log message. Currently it looks like
While this message requires a special format to comply with our testing (internal tools logstats.go and the E2E test), we can certainly trim it down and the UPDATED NextRound info is not crucial (it can be carved out as a DEBUG level message if not already present).
Problem
Some issues remaining at the end of #128. However, the long validation process made it safer to push fixes as followups
Remove the
*pImpl
parameter fromRetries()
inconduit/pipeline/common.go
Per @winder
Alternatively, just make
Retries
privateTrim the special end of round message
In the final EC2 test, the size of the log at catchup grew from 14.1 GB to 16.2 GB. This is attributable to the change of the end of round log message. Currently it looks like
In particular the
UPDATED...
prefix was added.While this message requires a special format to comply with our testing (internal tools
logstats.go
and the E2E test), we can certainly trim it down and the UPDATEDNextRound
info is not crucial (it can be carved out as aDEBUG
level message if not already present).Urgency
Medium