-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/51909/
-----------------------------------------------------------
Review request for mesos and Vinod Kone.
Repository: mesos
Description
-------
These benchmarks cover:
1. The time taken to make all registered agents unreachable and then
reachable. This is similar to what happens during a severe network
partition.
2. The time taken to GC a significant fraction (50%) of the
unreachable list from the registry.
Diffs
-----
src/tests/registrar_tests.cpp b04fc92c66449f1094212a534899e16d5f6f01bd
Diff: https://reviews.apache.org/r/51909/diff/
Testing
-------
make check
The performance of the new registry operations seems quite acceptable for
reasonable cluster sizes, despite the fact that `PruneUnreachable` is quadratic
time and `MarkSlaveReachable` and `MarkSlaveUnreachable` are both linear time.
For the time being, optimizing these operations doesn't seem like it is needed
urgently.
Thanks,
Neil Conway
