accumulo-notifications mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christopher Tubbs (JIRA)" <>
Subject [jira] [Commented] (ACCUMULO-1593) Looking up instanceName in ZK doesn't use consistent encoding
Date Thu, 05 Sep 2013 18:52:52 GMT


Christopher Tubbs commented on ACCUMULO-1593:

I think it makes sense to ensure the name and the id are both UTF-8 encoded when put in ZK.
Forcing this may break some fringe systems, but at the benefit of actually defining previously
undefined behavior. I think it's a reasonable risk.

As for the UUID being stored as bytes from the numbers... I don't think it's a big deal. More
importantly, though, there's a reason not to change it: Mike is right that it would affect
upgrades... but it might be worse than that... because we don't store version information
in the ZK mapping from name to instanceID. So, newer versions and older versions installed
simultaneously may conflict when trying to read this mapping.
> Looking up instanceName in ZK doesn't use consistent encoding
> -------------------------------------------------------------
>                 Key: ACCUMULO-1593
>                 URL:
>             Project: Accumulo
>          Issue Type: Bug
>            Reporter: Christopher Tubbs
>            Assignee: Ted Yu
>              Labels: newbie
>             Fix For: 1.6.0
>         Attachments: accumulo-1593-v1.txt, accumulo-1593-v2.txt
> We need to be more careful about encoding and decoding instance names and IDs in zookeeper.
They should use UTF-8 explicitly, right now, they use the JVM default, which can differ between
JVM runs (even between Initialize and start-all).

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:

View raw message