[
https://issues.apache.org/jira/browse/CASSANDRA-13580?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16042756#comment-16042756
]
Stefan Podkowinski commented on CASSANDRA-13580:
------------------------------------------------
If the scope was to simply exclude DCs from repairs, wouldn't {{repair -dc}}
have the same effect?
I'm also still trying to wrap my head around the help text of the new repair
option:
"Use --ignore_readonly_dcs to do a full sync on repair - even stream out data
from read only dcs."
What's being ignored here? The read only nodes from the repair or the read only
semantics that would usually be in place ignoring those nodes by default during
repairs? Does this option imply -full?
I'd personally avoid naming this "readonly datacenter", as people likely have
different expectations about the actual meaning. E.g. nodes in such a DC might
still be able to effect the cluster through actions like gossip, schema, or
read repairs. Nodes in a read only DC will still receive and persist data from
the rest of the clusters and get counted against the CL, right?
> Readonly datacenter support
> ---------------------------
>
> Key: CASSANDRA-13580
> URL: https://issues.apache.org/jira/browse/CASSANDRA-13580
> Project: Cassandra
> Issue Type: Bug
> Reporter: Marcus Eriksson
> Assignee: Marcus Eriksson
> Fix For: 4.x
>
>
> Some setups include datacenters where only reads are performed (example could
> be datacenter dedicated for taking backups).
> We could use this information during repair to make sure that we never stream
> out of a read only dc.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
