cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sylvain Lebresne (JIRA)" <>
Subject [jira] Commented: (CASSANDRA-1546) (Yet another) approach to counting
Date Wed, 20 Oct 2010 07:57:25 GMT


Sylvain Lebresne commented on CASSANDRA-1546:

bq. The limitation is during RESTORE_REPLICA_COUNT or BOOTSTRAP where a node that was previously
in the replica set becomes part of the replica set, again. In this case, when CCs are streamed
back to the, now re-added, host replica, it will discard those CCs that it should treat as

As you said, I think it is fine because of the fact the patches now use UUID as node ids.
The idea is that you'll never add back someone with 
the same node id ever. That is, if you boostrap a node, it'll affect a brand new, different
from every previously assigned node id (and maybe boostrap
should just start to remove the node id from the system table if it exists to force this regeneration).

If you add back a node previously removed, either the node still has its old sstables (and
so, for his node id, he has the correct count), and thus
you can add it back safely. Or you just force regenerating a new node id.

So I believe that it means that whenever you lose a sstable on a given node (for any reason),
you'll regenerate the node id. Now of course we 
should be careful to not regenerate the node id too often because it'll make the counter 'grow'.
But I doubt it will be a problem.

bq. Another note on #1546, the digest creation for CC and LCC do not create the same hashes.

True, that's the last bullet on my todo list (but my todo list may be lacking bullets). My
plan is to special case the digest creation indeed. 

> (Yet another) approach to counting
> ----------------------------------
>                 Key: CASSANDRA-1546
>                 URL:
>             Project: Cassandra
>          Issue Type: New Feature
>          Components: Core
>            Reporter: Sylvain Lebresne
>            Assignee: Sylvain Lebresne
>             Fix For: 0.7.1
>         Attachments: 0001-v2-Remove-IClock-from-internals.patch, 0001-v3-Remove-IClock-from-internals.txt,
0001-v4-Counters.patch, 0002-v2-Counters.patch, 0002-v3-Counters.txt, 0002-v4-thrift-changes.patch,
0003-v2-Thrift-changes.patch, 0003-v3-Thrift-changes.txt, marker_idea.txt
> This could be described as a mix between CASSANDRA-1072 without clocks and CASSANDRA-1421.
> More details in the comment below.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message