yugabyte-db
yugabyte-db copied to clipboard
[YSQL] Decreased Total ops after upgrade from 2.18.4.0-b52 to 2024.1 (INSERT only workloads)
Jira Link: DB-10940
Description
Please find slack thread in JIRA.
Steps: 7 nodes, RF3 universe setup. Started multiple INSERT only workloads with single read/write thread for all workloads, such that CPU comes to 70% or 75% range. Run for an hour on 2.18 and then upgrade to 2024.1
Observation: It hits https://github.com/yugabyte/yugabyte-db/issues/21979 but if databases are less (QA note: by increasing read/write threads to 3 or 5), it comes back after 3 to 4 mins. But, total ops, is much lesser and stays in the same level even after hours.
Workload Example:
java -jar /tmp/tests/artifacts/stress-sample-app-tool/yb-stress-sample-apps-1.1.43.jar --workload SqlDataLoad --num_writes -1 --num_threads_write 1 --num_threads_read 1 --num_reads -1 --num_unique_keys 100000000000000 --batch_size 3 --num_value_columns 10 --create_table_name table_db_28015f --default_postgres_database db_28015f --uuid_column --uuid 7a8a205e-51f2-447d-849e-3a2a593c5ecb --large_key_multiplier 3 --large_value_multiplier 3 --use_hikari true --uuid_marker db_28015f_file --nodes 172.151.23.72:5433,172.151.23.146:5433,172.151.26.159:5433,172.151.25.68:5433,172.151.18.7:5433,172.151.31.67:5433,172.151.17.128:5433
Read and write code can be seen here: https://github.com/yugabyte/yb-stress-test/blob/master/tools/sample-app/src/main/java/com/yugabyte/sample/apps/SqlDataLoad.java
Issue Type
kind/bug
Warning: Please confirm that this issue does not contain any sensitive information
- [X] I confirm this issue does not contain any sensitive information.