diff options
| author | Michael Klishin <mklishin@pivotal.io> | 2016-08-02 06:06:09 -0700 |
|---|---|---|
| committer | Michael Klishin <mklishin@pivotal.io> | 2016-08-02 06:06:09 -0700 |
| commit | 747b41a9d3794706390b21003eb4bf6929422213 (patch) | |
| tree | 0dac98f38ce913e31e06fc7deccf0fa69951acaf /scripts | |
| parent | 995c430fbb1981d81f8c18740153c611c49fda4d (diff) | |
| download | rabbitmq-server-git-747b41a9d3794706390b21003eb4bf6929422213.tar.gz | |
Switch to a single table per node for connection tracking
This way we have a single writer, multiple readers and no
lost counter update due to natural race conditions.
Data for tables from available nodes are then aggregated.
Two open questions:
* Whether to use a parallel version of lists:map/2
* When to clean up the tables
Diffstat (limited to 'scripts')
0 files changed, 0 insertions, 0 deletions