OracleRAC & ASM
Global cache (interconnect) waits by instance
Cluster-class waits per instance since startup. High average times on gc cr or gc current events point at the interconnect or at hot blocks shared across nodes.
Not yet verified. How scripts are tested
1CLEAR COLUMNS2SET LINESIZE 140 PAGESIZE 100 TRIMOUT ON TAB OFF3COLUMN inst_id HEADING "INST" FORMAT 9994COLUMN event FORMAT A405COLUMN total_waits HEADING "WAITS" FORMAT 999,999,999,9906COLUMN waited_s HEADING "TOTAL S" FORMAT 999,999,9907COLUMN avg_ms HEADING "AVG MS" FORMAT 99,990.998 9SELECT inst_id, event, total_waits,10 ROUND(time_waited_micro / 1e6) AS waited_s,11 ROUND(time_waited_micro / NULLIF(total_waits, 0) / 1000, 2) AS avg_ms12FROM gv$system_event13WHERE wait_class = 'Cluster'14 AND total_waits > 015ORDER BY time_waited_micro DESC;Save it as ora-rac-gc.sql and run it with SQL> @ora-rac-gc.
More Oracle scripts: RAC & ASM
- Stop, start and check a RAC database (srvctl)Clusterware commands to stop the database, start it read-only, check its status, and start a single instance in MOUNT mode. Run one line at a time;…
- ASM disk group spaceSize, free and usable space per disk group. Usable GB already accounts for mirroring, so it's the number to alert on. Works from the ASM or database…
- ASM disks and their statusEvery disk ASM can see, including CANDIDATE disks not yet in a group. Check MODE and STATE after storage maintenance.
- ASM rebalance progressRebalance operations in flight with power level and estimated minutes left. No rows means nothing is rebalancing.
- Sessions per instance and serviceHow connections are spread across RAC nodes and services. A lopsided spread usually means a service isn't running where you think it is.
- Cluster status commands (srvctl and crsctl)The Clusterware commands for checking a RAC database, its services and every cluster resource.