ESQL: Reserve memory TopN #134235
Merged
ESQL: Reserve memory TopN #134235
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking!
elasticsearchmachine
added
the
Team:Analytics
|
Pinging @elastic/es-analytical-engine (Team:Analytics) |
|
Hi @nik9000, I've created a changelog YAML for you. |
…topn' into esql_reserve_memory_for_lucene_topn
nik9000
changed the title
dnhatn
approved these changes
Sep 5, 2025
| this.encoders = encoders; | ||
| this.sortOrders = sortOrders; | ||
| this.inputQueue = new Queue(topCount); | ||
| breaker.addEstimateBytesAndMaybeBreak(Queue.sizeOf(topCount), "esql engine topn"); |
There was a problem hiding this comment.
nit: maybe move this memory acquire to Queue ctor for consistency with Queue#close.
There was a problem hiding this comment.
I'll leave a comment about why I'm not doing that! The trouble is that we allocate the memory in the super. I guess I can make the ctor private and call a static. that's good. I'll do that.
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
I committed it in elastic#134235 by accident. We were going to use it as part of that but decided against it.
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 8, 2025
I committed it in #134235 by accident. We were going to use it as part of that but decided against it.
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
to nik9000/elasticsearch
that referenced
this pull request
Sep 8, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
nik9000
added a commit
that referenced
this pull request
Sep 9, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
rjernst
pushed a commit
to rjernst/elasticsearch
that referenced
this pull request
Sep 9, 2025
I committed it in elastic#134235 by accident. We were going to use it as part of that but decided against it.
elasticsearchmachine
pushed a commit
that referenced
this pull request
Sep 9, 2025
* ESQL: Reserve memory TopN (#134235) Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big! * fix backport
sarog
pushed a commit
to portsbuild/elasticsearch
that referenced
this pull request
Sep 11, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
sarog
pushed a commit
to portsbuild/elasticsearch
that referenced
this pull request
Sep 19, 2025
Tracks the more memory that's involved in topn. Lucene doesn't track memory usage for TopN and can use a fair bit of it. Try this query: ``` FROM big_table | SORT a, b, c, d, e | LIMIT 1000000 | STATS MAX(a) ``` We attempt to return all million documents from lucene. Is we did this with the compute engine we're track all of the memory usage. With lucene we have to reserve it. In the case of the query above the sort keys weight 8 bytes each. 40 bytes total. Plus another 72 for Lucene's `FieldDoc`. And another 40 at least for copying to the values to `FieldDoc`. That totals something like 152 bytes a piece. That's 145mb. Worth tracking! ## Esql Engine TopN Esql *does* track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Tracks the more memory that's involved in topn.
Lucene TopN
Lucene doesn't track memory usage for TopN and can use a fair bit of it.
Try this query:
We attempt to return all million documents from lucene. Is we did this
with the compute engine we're track all of the memory usage. With lucene
we have to reserve it.
In the case of the query above the sort keys weight 8 bytes each. 40
bytes total. Plus another 72 for Lucene's
FieldDoc. And another 40 atleast for copying to the values to
FieldDoc. That totals somethinglike 152 bytes a piece. That's 145mb. Worth tracking!
Esql Engine TopN
Esql does track memory for topn, but it doesn't track the memory used by the min heap itself. It's just a big array of pointers. But it can get very big!