ORACLE 11g RAC — Recovering from an Interrupted Grid Install GUI and Completing the DB Engine Silent InstallRoot-causing INS-35354 · Repairing Central Inventory · CVU verification · DBCA prep — 2-node RAC lab notes (oel11db1 · oel11db2)#Oracle11gRAC#GridInfrastructure#INS-35354#CentralInventory#CVU#DBCA1. What Happened and the Bottom LineLate in the Grid Infrastructure install, root.sh had alrea..
Oracle 19c: Turning a Correlated IN That Looped 10,000 Times Into an EXISTS Semi JoinThe problematic, repeated partWhat changed after the fixWhen analyzing production SQL, it's common to find cases where changing IN to EXISTS made a query faster. But IN isn't always slow, and EXISTS isn't always fast — the Oracle Optimizer can transform both expressions into the same Semi Join.What matters isn't..
A manual Oracle install has a long tail of checkpoints — account creation, kernel parameters, udev rules, root scripts, RU patches, DBCA.This is a walkthrough of automating that entire path with Ansible, from a bare Rocky Linux box to a verified Oracle Restart (HAS) database — with the actual playbook output kept in, not summarized away.1. The GoalOracle installs are procedure-heavy by nature. A..
Sharable Memory north of 10GB.Thousands of INSERT ALL statements, identical in shape, differing only in literal values.Switching to bind variables — without touching the cursor lifecycle otherwise — changed both Shared Pool occupancy and Hard Parse counts completely. Confirmed against V$SQLAREA, not assumed.1. The ProblemA Shared Pool issue came in for review. Pulling up V$SQLAREA showed one SQL..
Soft Parse % sat at 99.98%.Execute to Parse % sat at 9.22%.Closing and reopening the same cursor on every call was the difference between a 0% and a 99.86% Execute to Parse ratio — confirmed against v$sqlstats, not assumed.1. The ProblemAn AWR report came in for review. Every Instance Efficiency metric cleared 99%:Buffer Hit %99.91Library Hit %99.94Soft Parse %99.98Latch Hit %99.97Execute to Par..
HASH JOIN ANTI was not the problem.The optimizer's join order forced TB_MATERIAL to be probed 12,380 times — that was the real issue.Changing the driving table alone cut Logical Reads from 38,350 to 7,412 — an 81% reduction.1. The ProblemA batch INSERT query populates a content business table based on contract data. The conditions looked straightforward: Filter by a specific corporation (ID_COR..
