Failed to create index from local gateway

Lukas_Vlcek1 · November 13, 2010, 9:54am

Hi,

I noticed interesting issue today. Not sure if I have the latest master but
it is not older then one day.

I created few indices in cluster, then I shut down the cluster and started
again. During startup I got the following ERROR message in log:

[ERROR][gateway.local ] [Drake, Frank] failed to create index
[testindex]

I shutdown the cluster and started again, this time I did not get this error
but the index was not recognized and created at all. It is simply missing in
cluster status. The index data is still present in /data folder but if I
check the _state folder there are two files: metadata-xx and shards-yy. The
[testindex] is still referenced in shards-yy but not in metadata-xx.

Not sure what went wrong (did not have a DEBUG logging level turned on) but
one of the other indices have alias [testindex]. So there is an index A and
index B having alias A, can this be a problem?

Regards,
Lukas

Lukas_Vlcek1 · November 13, 2010, 11:11am

I created issue for this (tested with master):

github.com/elastic/elasticsearch

Fail to create index when alias on other index

opened 11:10AM - 13 Nov 10 UTC

closed 11:16AM - 13 Nov 10 UTC

lukas-vlcek

Starting with clean ES installation: <pre> curl -XPOST http://localhost:9200/a…bc/x -d '{ "name" : "a" }' curl -XPOST http://localhost:9200/xyz/x -d '{ "name" : "b" }' curl -XPOST http://localhost:9200/_aliases -d '{ "actions" : [ {"add":{"index":"abc", "alias":"xyz"}} ] }' curl -XPOST http://localhost:9200/_shutdown </pre> then when starting ES again <pre> ./bin/elasticsearch -f </pre> I get the following log messages: <pre> [2010-11-13 12:03:24,159][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: initializing ... [2010-11-13 12:03:24,162][INFO ][plugins ] [Foster, Bill] loaded [] [2010-11-13 12:03:25,533][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: initialized [2010-11-13 12:03:25,533][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: starting ... [2010-11-13 12:03:25,685][INFO ][transport ] [Foster, Bill] bound_address {inet[/0:0:0:0:0:0:0:0:9300]}, publish_address {inet[/192.168.2.3:9300]} [2010-11-13 12:03:28,716][INFO ][cluster.service ] [Foster, Bill] new_master [Foster, Bill][R_kJChj2TtC1vyrqUK74QA][inet[/192.168.2.3:9300]], reason: zen-disco-join (elected_as_master) [2010-11-13 12:03:28,750][INFO ][discovery ] [Foster, Bill] elasticsearch/R_kJChj2TtC1vyrqUK74QA [2010-11-13 12:03:28,828][INFO ][http ] [Foster, Bill] bound_address {inet[/0:0:0:0:0:0:0:0:9200]}, publish_address {inet[/192.168.2.3:9200]} [2010-11-13 12:03:28,829][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: started [2010-11-13 12:03:29,444][INFO ][cluster.metadata ] [Foster, Bill] [abc] creating index, cause [gateway], shards [5]/[1], mappings [x] [2010-11-13 12:03:29,447][ERROR][gateway.local ] [Foster, Bill] failed to create index [xyz] [2010-11-13 12:03:29,914][INFO ][cluster.metadata ] [Foster, Bill] [abc] created and added to cluster_state </pre> Note the ERROR ^^ What is even more strange is that if I use index names [a] instead [abc] and [b] instead [xyx] then it works fine.

On Sat, Nov 13, 2010 at 10:54 AM, Lukáš Vlček lukas.vlcek@gmail.com wrote:

Hi,

I noticed interesting issue today. Not sure if I have the latest master but
it is not older then one day.

I created few indices in cluster, then I shut down the cluster and started
again. During startup I got the following ERROR message in log:

[ERROR][gateway.local ] [Drake, Frank] failed to create index
[testindex]

I shutdown the cluster and started again, this time I did not get this
error but the index was not recognized and created at all. It is simply
missing in cluster status. The index data is still present in /data folder
but if I check the _state folder there are two files: metadata-xx and
shards-yy. The [testindex] is still referenced in shards-yy but not in
metadata-xx.

Not sure what went wrong (did not have a DEBUG logging level turned on) but
one of the other indices have alias [testindex]. So there is an index A and
index B having alias A, can this be a problem?

Regards,
Lukas

Lukas_Vlcek1 · November 13, 2010, 11:15am

Sorry for creating the ticket twice:

github.com/elastic/elasticsearch

Fail to create index when alias on other index

opened 11:10AM - 13 Nov 10 UTC

closed 09:02PM - 13 Nov 10 UTC

lukas-vlcek

Starting with clean ES installation: <pre> curl -XPOST http://localhost:9200/a…bc/x -d '{ "name" : "a" }' curl -XPOST http://localhost:9200/xyz/x -d '{ "name" : "b" }' curl -XPOST http://localhost:9200/_aliases -d '{ "actions" : [ {"add":{"index":"abc", "alias":"xyz"}} ] }' curl -XPOST http://localhost:9200/_shutdown </pre> then when starting ES again <pre> ./bin/elasticsearch -f </pre> I get the following log messages: <pre> [2010-11-13 12:03:24,159][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: initializing ... [2010-11-13 12:03:24,162][INFO ][plugins ] [Foster, Bill] loaded [] [2010-11-13 12:03:25,533][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: initialized [2010-11-13 12:03:25,533][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: starting ... [2010-11-13 12:03:25,685][INFO ][transport ] [Foster, Bill] bound_address {inet[/0:0:0:0:0:0:0:0:9300]}, publish_address {inet[/192.168.2.3:9300]} [2010-11-13 12:03:28,716][INFO ][cluster.service ] [Foster, Bill] new_master [Foster, Bill][R_kJChj2TtC1vyrqUK74QA][inet[/192.168.2.3:9300]], reason: zen-disco-join (elected_as_master) [2010-11-13 12:03:28,750][INFO ][discovery ] [Foster, Bill] elasticsearch/R_kJChj2TtC1vyrqUK74QA [2010-11-13 12:03:28,828][INFO ][http ] [Foster, Bill] bound_address {inet[/0:0:0:0:0:0:0:0:9200]}, publish_address {inet[/192.168.2.3:9200]} [2010-11-13 12:03:28,829][INFO ][node ] [Foster, Bill] {elasticsearch/0.13.0-SNAPSHOT/2010-11-13T10:04:15}[8348]: started [2010-11-13 12:03:29,444][INFO ][cluster.metadata ] [Foster, Bill] [abc] creating index, cause [gateway], shards [5]/[1], mappings [x] [2010-11-13 12:03:29,447][ERROR][gateway.local ] [Foster, Bill] failed to create index [xyz] [2010-11-13 12:03:29,914][INFO ][cluster.metadata ] [Foster, Bill] [abc] created and added to cluster_state </pre> Note the ERROR ^^ What is even more strange is that if I use index names [a] instead [abc] and [b] instead [xyx] then it works fine.

https://github.com/elasticsearch/elasticsearch/issues/issue/512The
network is really weak here this morning. Feel free to kill one of these
tickets.

On Sat, Nov 13, 2010 at 12:11 PM, Lukáš Vlček lukas.vlcek@gmail.com wrote:

I created issue for this (tested with master):
Fail to create index when alias on other index · Issue #513 · elastic/elasticsearch · GitHub

On Sat, Nov 13, 2010 at 10:54 AM, Lukáš Vlček lukas.vlcek@gmail.comwrote:

Hi,

I noticed interesting issue today. Not sure if I have the latest master
but it is not older then one day.

I created few indices in cluster, then I shut down the cluster and started
again. During startup I got the following ERROR message in log:

[ERROR][gateway.local ] [Drake, Frank] failed to create index
[testindex]

I shutdown the cluster and started again, this time I did not get this
error but the index was not recognized and created at all. It is simply
missing in cluster status. The index data is still present in /data folder
but if I check the _state folder there are two files: metadata-xx and
shards-yy. The [testindex] is still referenced in shards-yy but not in
metadata-xx.

Not sure what went wrong (did not have a DEBUG logging level turned on)
but one of the other indices have alias [testindex]. So there is an index A
and index B having alias A, can this be a problem?

Regards,
Lukas

Lukas_Vlcek1 · November 13, 2010, 11:16am

I closed #513. Sorry for the confusion.

On Sat, Nov 13, 2010 at 12:15 PM, Lukáš Vlček lukas.vlcek@gmail.com wrote:

Sorry for creating the ticket twice:
Fail to create index when alias on other index · Issue #512 · elastic/elasticsearch · GitHub
https://github.com/elasticsearch/elasticsearch/issues/issue/512The
network is really weak here this morning. Feel free to kill one of these
tickets.

On Sat, Nov 13, 2010 at 12:11 PM, Lukáš Vlček lukas.vlcek@gmail.comwrote:

I created issue for this (tested with master):
Fail to create index when alias on other index · Issue #513 · elastic/elasticsearch · GitHub

On Sat, Nov 13, 2010 at 10:54 AM, Lukáš Vlček lukas.vlcek@gmail.comwrote:

Hi,

I noticed interesting issue today. Not sure if I have the latest master
but it is not older then one day.

I created few indices in cluster, then I shut down the cluster and
started again. During startup I got the following ERROR message in log:

[ERROR][gateway.local ] [Drake, Frank] failed to create index
[testindex]

I shutdown the cluster and started again, this time I did not get this
error but the index was not recognized and created at all. It is simply
missing in cluster status. The index data is still present in /data folder
but if I check the _state folder there are two files: metadata-xx and
shards-yy. The [testindex] is still referenced in shards-yy but not in
metadata-xx.

Not sure what went wrong (did not have a DEBUG logging level turned on)
but one of the other indices have alias [testindex]. So there is an index A
and index B having alias A, can this be a problem?

Regards,
Lukas