Does index.translog.disable_flush settings affect whole cluster or just specific node?


(Piavlo) #1

Hi,

I have a cluster in ec2 and I want to make cluster wide consistent
backup by:
1)set index.translog.disable_flush=true
2)make manual flush
3)snapshot elastic search datadir EBS volume on each node.

The question the index.translog.disable_flush and
http://localhost:9200/_flush are per node actions which i need to run on
each node or they are cluster wide and can be issued just once against
any node in the cluster?

Thanks
Alex


(Igor Motov) #2

Flush and update settings are per-index operations. If you will run them
without specifying index name (using /_flush and /_settings) on any node in
the cluster, they will run on all indices on all nodes.

On Sunday, April 29, 2012 6:30:04 AM UTC-4, Piavlo wrote:

Hi,

I have a cluster in ec2 and I want to make cluster wide consistent
backup by:
1)set index.translog.disable_flush=true
2)make manual flush
3)snapshot elastic search datadir EBS volume on each node.

The question the index.translog.disable_flush and
http://localhost:9200/_flush are per node actions which i need to run on
each node or they are cluster wide and can be issued just once against
any node in the cluster?

Thanks
Alex


(Piavlo) #3

Thanks for your reply Igor

And is there way to know if the /_flush has finished on all cluster nodes?

Thanks

On 04/29/2012 06:26 PM, Igor Motov wrote:

Flush and update settings are per-index operations. If you will run
them without specifying index name (using /_flush and /_settings) on
any node in the cluster, they will run on all indices on all nodes.

On Sunday, April 29, 2012 6:30:04 AM UTC-4, Piavlo wrote:

  Hi,

I have a cluster in ec2 and I want to make cluster wide consistent
backup by:
1)set index.translog.disable_flush=true
2)make manual flush
3)snapshot elastic search datadir EBS volume on each node.

The question the index.translog.disable_flush and
http://localhost:9200/_flush are per node actions which i need to
run on
each node or they are cluster wide and can be issued just once
against
any node in the cluster?


Thanks
Alex

(Igor Motov) #4

The REST _flush method is blocking. It returns when _flush on all shards
either finishes successfully or fails and returns corresponding numbers.

On Monday, April 30, 2012 11:31:27 AM UTC-4, Piavlo wrote:

Thanks for your reply Igor

And is there way to know if the /_flush has finished on all cluster nodes?

Thanks

On 04/29/2012 06:26 PM, Igor Motov wrote:

Flush and update settings are per-index operations. If you will run them
without specifying index name (using /_flush and /_settings) on any node in
the cluster, they will run on all indices on all nodes.

On Sunday, April 29, 2012 6:30:04 AM UTC-4, Piavlo wrote:

Hi,

I have a cluster in ec2 and I want to make cluster wide consistent
backup by:
1)set index.translog.disable_flush=true
2)make manual flush
3)snapshot elastic search datadir EBS volume on each node.

The question the index.translog.disable_flush and
http://localhost:9200/_flush are per node actions which i need to run on
each node or they are cluster wide and can be issued just once against
any node in the cluster?

Thanks
Alex


(system) #5