Database Reference
In-Depth Information
FAIRNESS_CLEARS NUMBER
FREE_GC_ELEMENTS NUMBER
FLUSHES NUMBER
FLUSHES_QUEUED NUMBER
FLUSH_QUEUE_FULL NUMBER
FLUSH_MAX_TIME NUMBER
LIGHT_WORKS NUMBER
ERRORS NUMBER
Script: MVRACPDnTap_crblksrvr.sql
SELECT inst_id INT,
( cr_requests + current_requests ) "BR",
( data_requests + undo_requests + tx_requests ) "HBLMS",
disk_read_results "DR",
fail_results "FR",
fairness_down_converts "FDC",
flushes,
flush_max_time "FMT",
light_works "LW",
errors "ERR"
FROM gv$cr_block_server
ORDER BY inst_id;
Reqsts
Blocks Handled Disk Down Flush
In Received By LMS Reads Failures Converts Flushes Time LW Errors
--- -------- -------- -------- -------- -------- -------- -------- ------ ------
1 47331 45948 2 5 14386 4789 198633 1098 0
2 83893 80222 720 16 34915 2776 253516 2134 0
3 60262 56767 373 24 18598 2963 126301 1999 0
4 65914 62605 484 5 19511 3013 100035 1932 0
5 81390 67136 340 14 36161 2392 117637 1154 0
6 1848995 1831623 860 610 1081569 99468 535471 32224 0
7 1690931 1677306 918 193 994073 140368 1468958 8544 0
8 2626341 2609583 701 229 1616399 107090 152438 11218 0
The preceding output gives the total consistent read requests received by the respective instances, the number
of consistent read blocks served by the LMS background process, and the number of failures during the activity.
However, the counts in the “failure” column are not due to lost blocks but exceptions created during the “prepare”
phase. In the preceding output, instances 6, 7, and 8 have significantly high activity because these instances run the
ETL (extract, transform, load) batch loads.
■
When there is high number of “failures,” it's important to follow up with oracle support to ensure there are no
known bugs. When creating service requests, the lMS trace files should be provided for analysis.
Note
Historical data for this view is maintained in
DBA_HIST_CR_BLOCK_SERVER.
Search WWH ::
Custom Search