You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@impala.apache.org by jo...@apache.org on 2023/06/02 16:21:20 UTC
[impala] 01/02: IMPALA-12142: Decrease default fetch_size to 8192 in impala-shell
This is an automated email from the ASF dual-hosted git repository.
joemcdonnell pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/impala.git
commit adfa6c83ec8ca371798b17021d49a8eacd954faf
Author: Csaba Ringhofer <cs...@cloudera.com>
AuthorDate: Thu Jun 1 11:52:23 2023 +0200
IMPALA-12142: Decrease default fetch_size to 8192 in impala-shell
The previous fetch_size of 10240 turned out to be suboptimal for HS2
server side, likely because it leads to overallocation in result
'std::vector's. Changed to the closest power of 2 size (8192).
With this change RowMaterializationTimer decreased from 3.4s to 2.7s
for "SELECT * FROM tpch_parquet.lineitem".
Change-Id: I34973cb705db53c496b9944c74995b45cf720d46
Reviewed-on: http://gerrit.cloudera.org:8080/19965
Reviewed-by: Kurt Deschler <kd...@cloudera.com>
Reviewed-by: Daniel Becker <da...@cloudera.com>
Tested-by: Impala Public Jenkins <im...@cloudera.com>
---
shell/option_parser.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/shell/option_parser.py b/shell/option_parser.py
index 75ede820a..1fc6fde2d 100755
--- a/shell/option_parser.py
+++ b/shell/option_parser.py
@@ -320,7 +320,7 @@ def get_option_parser(defaults):
"enforce any http path for the incoming requests, deployments could "
"still put it behind a loadbalancer that can expect the traffic at a "
"certain path.")
- parser.add_option("--fetch_size", type="int", dest="fetch_size", default=10240,
+ parser.add_option("--fetch_size", type="int", dest="fetch_size", default=8192,
help="The fetch size when fetching rows from the Impala coordinator. "
"The fetch size controls how many rows a single fetch RPC request "
"(RPC from the Impala shell to the Impala coordinator) reads at a "
@@ -328,7 +328,7 @@ def get_option_parser(defaults):
"('spool_query_results'=true). When result spooling is enabled "
"values over the batch_size are honored. When result spooling is "
"disabled, values over the batch_size have no affect. By default, "
- "the fetch_size is set to 10240 which is equivalent to 10 row "
+ "the fetch_size is set to 8192 which is equivalent to 8 row "
"batches (assuming the default batch size). Note that if result "
"spooling is disabled only a single row batch can be fetched at a "
"time regardless of the specified fetch_size.")