cassandra stats are broken if using order-preserving partioner

[clutchski]

We are parsing all Cassandra tokens as floats. This fails for order preserving partitioners (and perhaps others):

Random partitioners return a number for the token:

Token            : 113427455640312821154458202479064646084

However, Cassandra's order preserving partioner returns the following:

Token            : Token(bytes[56713727820156410577229101240436610842])

[clutchski]

This works for the order preserving partioner, but doesn't solve the byteorderedpartitioner, which returns a hash instead of a number. I think i'm going to just remove tracking the token entirely.

$ nodetool -h localhost info          
Token            : Token(bytes[17f0ab9fce7ad5f8afdfb64dad5a3537])

[clutchski]

Ping @alq666 in case you want to chime in.

[alq666]

Fine with me. The token is only marginally useful to graph, and even less
so as a time series.

On Thursday, September 27, 2012, Matt Perpick wrote:

Ping @alq666 https://github.com/alq666 in case you want to chime in.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/202#issuecomment-8951056.

*Alexis Lê-Quôc *+1-917-512-6452

Try Datadog for free (http://dtdg.co/alq-says-try-datadog)http://goo.gl/swO1h
no cc req'd
And follow @datadoghq https://twitter.com/#!/datadoghq on twitter