This commit allows code to be executed before or after a shards content
is deleted from disk. This is only executed if the shard owns the
content ie. on a shard file system only a primary shard will execute
these operations.
Several issues where reported showing truncated files where footers
didn't match and checksums read past EOF. This test reproduces the issue
on the latest 1.4 branch but passes on all versions above.
Closes#10155
Fixing geo_shape field mapper to persist the orientation parameter. Also adding parsing and integration tests to ensure persistence across cluster restarts.
Adds a setting to disable detailed error messages and full exception stack traces
in HTTP responses. When set to false, the error_trace request parameter will result
in a HTTP 400 response. When the error_trace parameter is not present, the message
of the first ElasticsearchException will be output and no nested exception messages
will be output.
Lines in the code that should be removed before a release can be annotated with
//NORELEASE . This can be useful when debugging test failures. For example,
one might want to add additional logging that would be too verbose for production
and therfore should be removed before releasing.
closes#10141
In case an exception was caught by the repeat rule, the retry mechanism would kick in only if the exception was the expected one. If not an NPE got thrown, while we should rather just bubble it up to the caller. This makes `NettyTransportMultiPortTests` run from a plane. An assumption would kick in to make sure that the test gets ignored but the `AssumptionViolationException` was caught and not properly re-thrown.
By default we won't allow rebalance operation if no all shards are active.
if this is the case we don't need to worry about costly rebalance calculations at all.
to enable download servers to send the correct plugin version for the
node that is installing it this PR sends the current version as a header
to the server.
In case a HTTP client connects to the transport protocol and issues a
HTTP method followed by a space, we can just try to be smart and return
a string back to the client to point the user to the fact that the wrong
port has been used.
Closes#2139Closes#10108
This commit adds the current total number of translog operations to the recovery reporting API. We also expose the recovered / total percentage:
```
"translog": {
"recovered": 536,
"total": 986,
"percent": "54.3%",
"total_time": "2ms",
"total_time_in_millis": 2
},
```
Closes#9368Closes#10042
The filter from an indexed alias is as if you would filter on the metadata of a percolator query, but then the filter is defined in the index alias instead of the percolate request.
Closes#6241
The behaviour is better in the case someone has multiple levels of nested object fields defined in the mapping and like to define a single inner_hits definition that is two or more levels deep.
If someone wants inner hits on a nested field that is 2 levels deep the following would need to be defined:
```
{
...
"inner_hits" : {
"path" : {
"level1" : {
"inner_hits" : {
"path" : {
"level2" : {
"query" : { .... }
}
}
}
}
}
}
}
```
With this change the above can be defined as:
```
{
...
"inner_hits" : {
"path" : {
"level1.level2" : {
"query" : { .... }
}
}
}
}
```
Closes#9251
The BWC tests also run against a snapshot build of previous release branches. Upon a failure it's important to know what commit exactly was used.
Closes#10111
PageCacheRecycler was mistakenly using the maximum number of items per page
instead of the byte size of a page. This could result in higher memory usage
than configured.
Close#10077
This solves a problem in the time zone rounding classes where time dates that
fall into a DST gap will cause joda time library to throw an exception.
Changing the conversion methods 'strict' option to false prevents this.
Closes#10025
To support real time gets, the engine keeps an in-memory map of recently index docs and their location in the translog. This is needed until documents become visible in Lucene. With 1.3.0, we have improved this map and made tightly integrated with refresh cycles in Lucene in order to keep the memory signature to a bare minimum. On top of that, if the version map grows above a 25% of the index buffer size, we proactively refresh in order to be able to trim the version map back to 0 (see #6363) . In the same release, we have fixed an issue where an update to the indexing buffer could result in an unwanted exception during recovery (#6667) . We solved this by waiting with updating the size of the index buffer until the shard was fully recovered. Sadly this two together can have a negative impact on the speed of translog recovery.
During the second phase of recovery we replay all operations that happened on the shard during the first phase of copying files. In parallel we start indexing new documents into the new created shard. At some point (phase 3 in the recovery), the translog replay starts to send operation which have already been indexed into the shard. The version map is crucial in being able to quickly detect this and skip the replayed operations, without hitting lucene. Sadly #6667 (only updating the index memory buffer once shard is started) means that a shard is using the default 64MB for it's index buffer, and thus only 16MB (25%) for the version map. This much less then the default index buffer size 10% of machine memory (shared between shards).
Since we don't flush anymore when updating the memory buffer, we can remove #6667 and update recovering shards as well. Also, we make the version map max size configurable, with the same default of 25% of the current index buffer.
Closes#10046
