[
https://issues.apache.org/jira/browse/SOLR-5656?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14056934#comment-14056934
]
Gregory Chanan commented on SOLR-5656:
--------------------------------------
bq. I figure this will be reused in a few other places that have to choose
nodes given a clusterstate.
I definitely think this sort of thing is useful, I just find it difficult to
parse currently
bq. The * marks a replica as the one being replaced. The current replacement
algorithm looks at each replica - when it finds one, it looks for the best
place to replace it given a clusterstate.
What I mean here is, for example, csr1-2*r2, does the * bind to (r1-2) or (r2).
bq. Currently it defaults to 1. I was going to make it explicit, but not a lot
of error checking yet anyway, so left it for further improvement later. I
figure this will be reused in a few other places that have to choose nodes
given a clusterstate.
I think making it explicit is a good idea. If being explicit isn't required,
is csr-2 legal?
bq. It just overrides a specific state for a replica in clusterstate.json - so
rather than ACTIVE, you could mark them as RECOVERING or DOWN.
I wonder if it would be clearer if these states were non-intersecting with the
node specification. So like, A=active, D=down, F=failed, R=recovering or if
you are worried about case, could make recovering C or M (for moving, there
isn't really a difference from a replica being moved vs recovering logically I
think)? Then you wouldn't need the minus either. A default of "A" there seems
reasonable too. What do you think?
> Add autoAddReplicas feature for shared file systems.
> ----------------------------------------------------
>
> Key: SOLR-5656
> URL: https://issues.apache.org/jira/browse/SOLR-5656
> Project: Solr
> Issue Type: New Feature
> Reporter: Mark Miller
> Assignee: Mark Miller
> Attachments: SOLR-5656.patch, SOLR-5656.patch, SOLR-5656.patch,
> SOLR-5656.patch
>
>
> When using HDFS, the Overseer should have the ability to reassign the cores
> from failed nodes to running nodes.
> Given that the index and transaction logs are in hdfs, it's simple for
> surviving hardware to take over serving cores for failed hardware.
> There are some tricky issues around having the Overseer handle this for you,
> but seems a simple first pass is not too difficult.
> This will add another alternative to replicating both with hdfs and solr.
> It shouldn't be specific to hdfs, and would be an option for any shared file
> system Solr supports.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
