Failure during refresh-mapping 0.90.5

Elastic Stack Elasticsearch
1

Hi,

After an upgrade to 90.5 we are seeing some issues that we find concerning
in the logs.

[2013-11-06 13:49:54,587][WARN ][cluster.metadata] [NODENAME] failure
during [refresh-mapping [INDEXNAME][[TYPENAMES]]]
java.lang.NullPointerException

I have browsed the code but can't at least in master see where this issue
might occur exactly. I am wondering what the issue might be and if someone
else have seen this.

Best Regards

Marcus Granström

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

2

hey can you try to figure out where this is coming from or maybe provide a
recreation?

simon

On Wednesday, November 6, 2013 5:10:32 PM UTC+1, Marcus Granstrom wrote:

Hi,

After an upgrade to 90.5 we are seeing some issues that we find concerning
in the logs.

[2013-11-06 13:49:54,587][WARN ][cluster.metadata] [NODENAME] failure
during [refresh-mapping [INDEXNAME][[TYPENAMES]]]
java.lang.NullPointerException

I have browsed the code but can't at least in master see where this issue
might occur exactly. I am wondering what the issue might be and if someone
else have seen this.

Best Regards

Marcus Granström

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

3

It might be related to:

That also happens when refresing mappings (the exact error message has
changed between 0.19. and 0.90)

On Wednesday, 6 November 2013 17:59:47 UTC+1, simonw wrote:

hey can you try to figure out where this is coming from or maybe provide a
recreation?

simon

On Wednesday, November 6, 2013 5:10:32 PM UTC+1, Marcus Granstrom wrote:

Hi,

After an upgrade to 90.5 we are seeing some issues that we find
concerning in the logs.

[2013-11-06 13:49:54,587][WARN ][cluster.metadata] [NODENAME] failure
during [refresh-mapping [INDEXNAME][[TYPENAMES]]]
java.lang.NullPointerException

I have browsed the code but can't at least in master see where this issue
might occur exactly. I am wondering what the issue might be and if someone
else have seen this.

Best Regards

Marcus Granström

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

4

When processing the mappings I can see from the debug output that the
parsed differs from the original. When looking in the source I then see:

IndicesClusterStateService.java:412
// we don't apply default, since it has been applied when the mappings were
parsed initially
mapperService.merge(mappingType, mappingSource.string(), false);
if
(!mapperService.documentMapper(mappingType).mappingSource().equals(mappingSource))
{
requiresRefresh = true;
// this might happen when upgrading from 0.15 to 0.16
logger.debug("[{}] parsed mapping [{}], and got different
sources\noriginal:\n{}\nparsed:\n{}", index, mappingType, mappingSource,
mapperService.documentMapper(mappingType).mappingSource());
}

Is that comment correct? (In my interpretation this should only happen when
upgrading from 0.15 to 0.16)
We actually see this happen in 0.90.5 when we have created the index in
0.90.5.

Another strange thing is that even though the mappings are processed for
each cluster change we don't get the error every time for a specific index
and type.

On Thursday, 7 November 2013 08:41:10 UTC+1, Henrik Lindström wrote:

It might be related to:
NullPointerException failed to dynamically refresh the mapping in cluster_state from shard · Issue #2633 · elastic/elasticsearch · GitHub

That also happens when refresing mappings (the exact error message has
changed between 0.19. and 0.90)

On Wednesday, 6 November 2013 17:59:47 UTC+1, simonw wrote:

hey can you try to figure out where this is coming from or maybe provide
a recreation?

simon

On Wednesday, November 6, 2013 5:10:32 PM UTC+1, Marcus Granstrom wrote:

Hi,

After an upgrade to 90.5 we are seeing some issues that we find
concerning in the logs.

[2013-11-06 13:49:54,587][WARN ][cluster.metadata] [NODENAME] failure
during [refresh-mapping [INDEXNAME][[TYPENAMES]]]
java.lang.NullPointerException

I have browsed the code but can't at least in master see where this
issue might occur exactly. I am wondering what the issue might be and if
someone else have seen this.

Best Regards

Marcus Granström

--
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 elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.