Unassigned shards after launching new node and then taking it down

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

Unassigned shards after launching new node and then taking it down

Mike
I have 1 node with 2 shards, 0 replicas, and a few indices. I'm able to access all of my data.
              nodeA
indexA    [0,1]
indexB    [0,1]
indexC    [0,1]
indexD    [0,1]


I brought up a 2nd node for testing, and the shards were distributed as such below.  I'm still able to access all of my data, which is good.
              nodeA     nodeB
indexA                  [0,1]
indexB    [0,1]
indexC    [0,1]
indexD                  [0,1]


I took down my second testing node, but now I can't access all my data like I had before.  Is this correct?  I thought my shards would have been brought back up on nodeA, similar to how they were moved to nodeB when it was brought up.
              nodeA     UNASSIGNED
indexA                  [0,1]
indexB    [0,1]
indexC    [0,1]
indexD                  [0,1]


1) How do I fix this?  Is the only solution now to bring back up my temporary nodeB, set replicas = 1 for all the indices, and then bring back down nodeB?  
2) To ensure maximum availability, is the recommendation to set # shards >= # expected nodes and # replicas == # expected nodes - 1?

--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/groups/opt_out.
 
 
Reply | Threaded
Open this post in threaded view
|

Re: Unassigned shards after launching new node and then taking it down

Clinton Gormley-2
Hiya Mike

On Thu, 2013-03-07 at 10:12 -0800, Mike wrote:
> I have 1 node with 2 shards, 0 replicas, and a few indices. I'm able
> to access all of my data.

> I took down my second testing node, but now I can't access all my data
> like I had before.  Is this correct?  

Yes - it relies on having replicas in order to be able to upgrade
"replica" shards to "primary" shards.  That's why it will never put a
primary and its replica on the same node.

> I thought my shards would have been brought back up on nodeA, similar
> to how they were moved to nodeB when it was brought up.
>               nodeA     UNASSIGNED
> indexA                  [0,1]
> indexB    [0,1]
> indexC    [0,1]
> indexD                  [0,1]
>
>
> 1) How do I fix this?  Is the only solution now to bring back up my
> temporary nodeB, set replicas = 1 for all the indices, and then bring
> back down nodeB?  

yes

> 2) To ensure maximum availability, is the recommendation to set #
> shards >= # expected nodes and # replicas == # expected nodes - 1?

yes

clint


--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/groups/opt_out.