IndexShardGatewayRecoveryException - but http turned off.


(davrob) #1

Hi,

I have the following exception (ElasticSearch 0.17.6):

IndexShardGatewayRecoveryException: [contactindex_2012_05_02_01_43_40][0]
shard allocated for local recovery (post api), should exists, but doesn't

For which, the accepted solution seems to be outlined here:
http://elasticsearch-users.115913.n3.nabble.com/Gateway-recovery-exception-tp3826325p3853899.html

Unfortunately, my http server is off by default, for security reasons, and
I can not connect using the Java client.

My config file, uses standard local gateway settings (i.e. whatever the
default is) as you can see here: https://gist.github.com/2636246

I would like to delete the local gateway to solve this issue,
unfortunately, I can't track where the local gateway files are stored. I
thought the default was {path.home}/work - but appears not to be the case.

-David.


(Shay Banon) #2

The local gateway data is stored under _state within the data location in
0.17.6. Any chance you can upgrade to 0.19? (you will first need to upgrade
to 0.18).

On Tue, May 8, 2012 at 6:26 PM, davrob2 daviroberts@gmail.com wrote:

Hi,

I have the following exception (ElasticSearch 0.17.6):

IndexShardGatewayRecoveryException: [contactindex_2012_05_02_01_43_40][0]
shard allocated for local recovery (post api), should exists, but doesn't

For which, the accepted solution seems to be outlined here:
http://elasticsearch-users.115913.n3.nabble.com/Gateway-recovery-exception-tp3826325p3853899.html

Unfortunately, my http server is off by default, for security reasons, and
I can not connect using the Java client.

My config file, uses standard local gateway settings (i.e. whatever the
default is) as you can see here: https://gist.github.com/2636246

I would like to delete the local gateway to solve this issue,
unfortunately, I can't track where the local gateway files are stored. I
thought the default was {path.home}/work - but appears not to be the case.

-David.


(davrob) #3

Hi Shay,

Actually all files under the data directory (data:
/app/fao/cmt/indexfolder/ES1) were deleted, though I can't check this for
myself. Should that have solved the issue?

An upgrade to 0.19 would not be impossible, but clearly its not an ideal
solution.

Just to be clear, are my options:

i) Persuade the guardians of my server to change http: enabled: false to
true, in which case, even though the server is in a 'dodgy' state it will
listen to a DELETE instruction.

ii) Upgrade to 0.19, which will ignore the error.

-David.

On Tuesday, 8 May 2012 16:43:37 UTC+1, kimchy wrote:

The local gateway data is stored under _state within the data location in
0.17.6. Any chance you can upgrade to 0.19? (you will first need to upgrade
to 0.18).

On Tue, May 8, 2012 at 6:26 PM, davrob2 <****> wrote:

Hi,

I have the following exception (ElasticSearch 0.17.6):

IndexShardGatewayRecoveryException: [contactindex_2012_05_02_01_43_40][0]
shard allocated for local recovery (post api), should exists, but doesn't

For which, the accepted solution seems to be outlined here:
http://elasticsearch-users.115913.n3.nabble.com/Gateway-recovery-exception-tp3826325p3853899.html

Unfortunately, my http server is off by default, for security reasons,
and I can not connect using the Java client.

My config file, uses standard local gateway settings (i.e. whatever the
default is) as you can see here: https://gist.github.com/2636246

I would like to delete the local gateway to solve this issue,
unfortunately, I can't track where the local gateway files are stored. I
thought the default was {path.home}/work - but appears not to be the case.

-David.


(Shay Banon) #4

If the root data location was deleted, then you won't get this problem (as
it has nothing to recover from). Upgrading to 0.19 will not help with this
error, it will help prevent this error.

On Tue, May 8, 2012 at 6:56 PM, davrob2 daviroberts@gmail.com wrote:

Hi Shay,

Actually all files under the data directory (data:
/app/fao/cmt/indexfolder/ES1) were deleted, though I can't check this for
myself. Should that have solved the issue?

An upgrade to 0.19 would not be impossible, but clearly its not an ideal
solution.

Just to be clear, are my options:

i) Persuade the guardians of my server to change http: enabled: false to
true, in which case, even though the server is in a 'dodgy' state it will
listen to a DELETE instruction.

ii) Upgrade to 0.19, which will ignore the error.

-David.

On Tuesday, 8 May 2012 16:43:37 UTC+1, kimchy wrote:

The local gateway data is stored under _state within the data location in
0.17.6. Any chance you can upgrade to 0.19? (you will first need to upgrade
to 0.18).

On Tue, May 8, 2012 at 6:26 PM, davrob2 <****> wrote:

Hi,

I have the following exception (ElasticSearch 0.17.6):

IndexShardGatewayRecoveryExcep**tion: [contactindex_2012_05_02_01_**43_40][0]
shard allocated for local recovery (post api), should exists, but doesn't

For which, the accepted solution seems to be outlined here:
http://elasticsearch-users.115913.n3.nabble.com/Gateway-
recovery-exception-**tp3826325p3853899.htmlhttp://elasticsearch-users.115913.n3.nabble.com/Gateway-recovery-exception-tp3826325p3853899.html

Unfortunately, my http server is off by default, for security reasons,
and I can not connect using the Java client.

My config file, uses standard local gateway settings (i.e. whatever the
default is) as you can see here: https://gist.github.com/**2636246https://gist.github.com/2636246

I would like to delete the local gateway to solve this issue,
unfortunately, I can't track where the local gateway files are stored. I
thought the default was {path.home}/work - but appears not to be the case.

-David.


(system) #5