[
https://issues.apache.org/jira/browse/CASSANDRA-9692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15154385#comment-15154385
]
Benedict commented on CASSANDRA-9692:
-------------------------------------
Hi [~Giampaolo],
2-4, feel free to address in this ticket - and not just those; in your link 4,
the transfer _rate_ doesn't need to be MB/s; in some cases KiB/s is probably
better, and also we probably mean MiB/s not MB/s anyway. Preferably all units
of all messages should be assessed and made better / dynamic.
Regarding 1, there is an advantage to capping the lower bound at KiB: it is
consistent for parsing, and hopefully makes it easier to spot where the value
they are looking for is, and immediately see its scale. Also, it is not at all
difficult to interpret the meaning of 0.099KiB, although I suppose the rounding
error might confuse people if they want to find a file with exactly that size.
Personally, I prefer the consistency of sticking with KiB, but that's obvious
since I wrote the method in the first place :)
> Print sensible units for all log messages
> -----------------------------------------
>
> Key: CASSANDRA-9692
> URL: https://issues.apache.org/jira/browse/CASSANDRA-9692
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Benedict
> Priority: Minor
> Labels: lhf
> Fix For: 3.x
>
>
> Like CASSANDRA-9691, this has bugged me too long. it also adversely impacts
> log analysis. I've introduced some improvements to the bits I touched for
> CASSANDRA-9681, but we should do this across the codebase. It's a small
> investment for a lot of long term clarity in the logs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
