[
https://issues.apache.org/jira/browse/CASSANDRA-1683?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12926480#action_12926480
]
Gary Dusbabek commented on CASSANDRA-1683:
------------------------------------------
bq. This isn't bootstrapping though, this is saying "join the cluster
immediately even though you don't have any data for the range you're claiming
responsibility for."
which is exactly what is implied when your turn autobootstrap off and boot a
new node. I didn't know we were discouraging this. If we are, the right thing
to do is size the data directories and throw a fatal error on startup if there
are no data files and autobootstrap is off.
bq. Right, but that's a subset of what I'm claiming: no node should serve reads
if it doesn't have the data it's claiming it has.
Is it also true that no node should serve writes if it doesn't have the data
it's claiming it has? Because we'd also be enforcing that.
> Bootstrapped node (autobootstrap==false) starts receiving writes before
> schema is propagated.
> ---------------------------------------------------------------------------------------------
>
> Key: CASSANDRA-1683
> URL: https://issues.apache.org/jira/browse/CASSANDRA-1683
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Affects Versions: 0.7 beta 1
> Reporter: Gary Dusbabek
> Assignee: Gary Dusbabek
> Fix For: 0.7.0
>
> Attachments:
> v2-0001-non-bootstrap-nodes-wait-for-schema-propagation.txt
>
>
> This is what generates the
> {code}org.apache.cassandra.db.UnserializableColumnFamilyException: Couldn't
> find cfId=1016{code} errors.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
