Testing Scripts and Leader Detection

[hcoffey1]

Mostly testing scripts, with an adjustment to the client RPC to search for the leader instead of relying on server direction.

[ajx42]

Can you try renaming the .log files to have .log.dat suffix, that should prevent GITHUB from counting those lines in the diff? Otherwise this change might pollute the diff especially when using commandline git.

[ajx42]

I have only looked at the C++ stuff. The scripts should we fine, we trust ye!

[ajx42]

I think we found that issue with 5 replicas and 1 failure with this thread join logic. Do you think we should still commit this (thread join) change?

[hcoffey1]

The performance benefit is nice, but since it does cause some issues, I've put it back to the 50 ms delay

[hcoffey1]

Can you try renaming the .log files to have .log.dat suffix, that should prevent GITHUB from counting those lines in the diff? Otherwise this change might pollute the diff especially when using commandline git.

Sure, I'll go ahead and adjust them

[hcoffey1]

Can you try renaming the .log files to have .log.dat suffix, that should prevent GITHUB from counting those lines in the diff? Otherwise this change might pollute the diff especially when using commandline git.

If its still an issue, I can also remove these files since they from when we were using the join approach for the leader iterations.

[ajx42]

looks okay to me