I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene not
remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang mock2u@gmail.com wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene not
remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang <moc...@gmail.com <javascript:>>
wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene
not remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang moc...@gmail.com wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene
not remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang <moc...@gmail.com <javascript:>>
wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene
not remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang moc...@gmail.com wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene
not remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
This only means that segment merging is going on. This produces generations
of segments that are obsolete, these are internally marked as "deleted".
Later the obsolete generations are cleaned up automatically when the
segment is touched by significant amount of new data. Do not worry about
that, there is nothing you should do. Of course, you can play with
_optimize API, but the downside is imposing heavy load on the node.
Jörg
On Tue, Feb 3, 2015 at 8:26 AM, Jason Zhang mock2u@gmail.com wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's
several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene not
remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
On 3 February 2015 at 18:26, Jason Zhang moc...@gmail.com wrote:
Hi,
I use segmentspy to detect segment merges in ES. But I find there's several "Deleted Docs" which is like "Deleted Docs: 1.55630".
I know ES will mark items as deleted and remove them later through GC.
But what's the meaning of those deleted segments? Why does ES/lucene not remove them? There's no updates in this index since 10 days ago.
How can I clean them out explicitly?
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.