[improvement](spill) avoid spill if memory is enough #33075
Merged
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
|
Thank you for your contribution to Apache Doris. Since 2024-03-18, the Document has been moved to doris-website. |
Contributor
Author
|
run buildall |
Contributor
|
clang-tidy review says "All clean, LGTM! 👍" |
|
TeamCity be ut coverage result: |
TPC-H: Total hot run time: 38742 ms |
TPC-DS: Total hot run time: 181873 ms |
ClickBench: Total hot run time: 30.6 s |
|
Load test result on machine: 'aliyun_ecs.c7a.8xlarge_32C64G' |
yiguolei
reviewed
Mar 31, 2024
| local_state._mem_tracker->set_consumption( | ||
| local_state._shared_state->in_mem_shared_state->sorter->data_size()); | ||
| if (eos) { | ||
| if (_enable_spill) { |
Contributor
There was a problem hiding this comment.
I think spill_sort_sink_operator is used when _enable_spill== true. So that I think this variable is always true in this case??
Contributor
Author
There was a problem hiding this comment.
We currently only support spill for full sort, but when creating the sort operatorX, it's hard to decide whether it's full sort.
It's in the SortSinkOperatorX::prepare phase to decide whether to use full sort, so we create
SpillSortSinkOperatorX when enable_sort_spill=true for all type of sort, but actually only enable spillable sort for full sort.
yiguolei
reviewed
Mar 31, 2024
| auto& local_state = get_local_state(state); | ||
| Defer defer([&] { | ||
| if (*eos) { | ||
| LOG(INFO) << "hash probe " << id() << " eos"; |
yiguolei
reviewed
Mar 31, 2024
| RETURN_IF_ERROR(revoke_memory(state)); | ||
| local_state.profile()->add_info_string( | ||
| "Spilled", local_state._shared_state->is_spilled ? "true" : "false"); | ||
| LOG(INFO) << "sort node " << id() |
yiguolei
reviewed
Mar 31, 2024
|
|
||
| RETURN_IF_ERROR(local_state.initiate_merge_spill_partition_agg_data(state)); | ||
| if (local_state._shared_state->is_spilled) { | ||
| RETURN_IF_ERROR(local_state.initiate_merge_spill_partition_agg_data(state)); |
Contributor
There was a problem hiding this comment.
This method will merge block again. Do we need clear the previous partition data??
Contributor
Author
There was a problem hiding this comment.
initiate_merge_spill_partition_agg_data will check whether the current partition still has data, if it is, it continue to output data from the current partition; if not, will merge the next partition.
yiguolei
reviewed
Mar 31, 2024
|
|
||
| RETURN_IF_ERROR(local_state.initiate_merge_spill_partition_agg_data(state)); | ||
| if (local_state._shared_state->is_spilled) { | ||
| RETURN_IF_ERROR(local_state.initiate_merge_spill_partition_agg_data(state)); |
Contributor
There was a problem hiding this comment.
why call init merge every time?? It is only called the first time or meet previous partition's EOS?
yiguolei
reviewed
Mar 31, 2024
| RETURN_IF_ERROR(partition->finish_current_spilling(eos)); | ||
| local_state.profile()->add_info_string( | ||
| "Spilled", local_state._shared_state->is_spilled ? "true" : "false"); | ||
| LOG(INFO) << "agg node " << id() |
Contributor
Author
There was a problem hiding this comment.
If profile is not enabled, we can seach log to check if query is spilled.
yiguolei
reviewed
Mar 31, 2024
| } else { | ||
| for (auto& partition : local_state._shared_state->spill_partitions) { | ||
| RETURN_IF_ERROR(partition->finish_current_spilling(eos)); | ||
| local_state.profile()->add_info_string( |
Contributor
There was a problem hiding this comment.
Not meet EOS.
Shoud add this to profile when the spilled is set to true at first time.
Because we will provide realtime profile in the future, user will find his query is spilled to disk.
If we set it when eos, then user will not find ths spill task when the query not spilled finished.
jacktengg
force-pushed
the
improve-spill
branch
from
877e5f6 to
6638112
Compare
Contributor
|
clang-tidy review says "All clean, LGTM! 👍" |
Contributor
Author
|
run buildall |
Contributor
|
clang-tidy review says "All clean, LGTM! 👍" |
|
TeamCity be ut coverage result: |
TPC-H: Total hot run time: 38791 ms |
TPC-DS: Total hot run time: 181785 ms |
ClickBench: Total hot run time: 30.56 s |
|
Load test result on machine: 'aliyun_ecs.c7a.8xlarge_32C64G' |
github-actions
bot
added
the
approved
Contributor
|
PR approved by at least one committer and no changes requested. |
Contributor
|
PR approved by anyone and no changes requested. |
yiguolei
pushed a commit
that referenced
this pull request
Apr 10, 2024
yiguolei
pushed a commit
that referenced
this pull request
Apr 10, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Proposed changes
Issue Number: close #xxx
Further comments
If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...