Dangling Index recovery and config


(Mark D) #1

Hi,

I have a two fold question. We are getting dangling indexes on a three
machine, three node setup, and I dont understand why they are getting
"dangled" I originally had this configuration, periodically I would
get a dangling index. There are two indexes with a total of 20K docs.

gateway:
type: fs
fs:
location: /var/lib/elasticsearch/gateway

index:
number_of_shards: 2
number_of_replicas: 1

I added this configuration thinking it would help, and it seemed to
recover the indexes ok for a while, but now it is not

recover_after_nodes: 3
recover_after_time: 5m
expected_nodes: 3

  1. Is there a way to recover a dangling index in this scenario.
  2. Any ideas why this would be happening with this configuration?

Thanks in advance.

Mark


(Shay Banon) #2

Why are you using shared FS gateway, is that a location shared by the 3
nodes? Just use the (default) local gateway.

On Wed, Jun 13, 2012 at 11:44 PM, Mark D mark.a.donnelly@gmail.com wrote:

Hi,

I have a two fold question. We are getting dangling indexes on a three
machine, three node setup, and I dont understand why they are getting
"dangled" I originally had this configuration, periodically I would
get a dangling index. There are two indexes with a total of 20K docs.

gateway:
type: fs
fs:
location: /var/lib/elasticsearch/gateway

index:
number_of_shards: 2
number_of_replicas: 1

I added this configuration thinking it would help, and it seemed to
recover the indexes ok for a while, but now it is not

recover_after_nodes: 3
recover_after_time: 5m
expected_nodes: 3

  1. Is there a way to recover a dangling index in this scenario.
  2. Any ideas why this would be happening with this configuration?

Thanks in advance.

Mark


(system) #3