rac节点一在年后出现挂起
本帖最后由 ray 于 2014-4-1 09:28 编辑rac节点一在年后出现几次挂起,通过sqlplus链接数据库出现排队等待不能显示sql>符号。在os级别kill一下oracle的链接进程问题解决。但是不能确定错误根源。请飞总指导。我上传了2个节点在挂起时的awr和ganlia的监控信息。
webkit-fake-url://52F6012E-8AB0-453C-B6C7-E6F36753AB1B/image.tiff
webkit-fake-url://6FE3A95D-A3B0-4E6E-95B2-5D659873444D/image.tiff
我看了awr
1节点的负载非常高已经超过了系统理论承受的范围
Snap IdSnap TimeSessionsCursors/Session
Begin Snap:499431-3月 -14 23:00:161213.1
End Snap:499531-3月 -14 23:47:071262.8
Elapsed: 46.85 (mins)
DB Time: 573.80 (mins)
StatisticTotal
PHYSICAL_MEMORY_BYTES33,705,668,608
NUM_CPUS8
NUM_CPU_SOCKETS2
考虑在系统参数,数据库参数,数据库bug,sql 优化,这三个方面着手,优化节点1,应该改故障就可以消除
Pool Name Begin MB End MB % Diff
java free memory 14.49 14.49 0.00
java joxlod exec hp 16.87 16.87 0.00
java joxs heap 0.65 0.65 0.00
large ASM map operations hashta 0.38 0.38 0.00
large PX msg pool 1.03 1.03 0.00
large free memory 14.60 14.60 0.00
shared CCursor 97.46 43.50 -55.37
shared KGH: NO ACCESS 244.91 610.41 149.24
shared KQR L PO 108.47 -100.00
shared PCursor 53.92 35.79 -33.62
shared db_block_hash_buckets 90.00 90.00 0.00
shared free memory 440.87 545.38 23.71
shared gcs resources 361.48 361.48 0.00
shared gcs shadows 216.44 216.44 0.00
shared ges resource 50.74 35.92 -29.22
shared kglsim object batch 24.22 24.22 0.00
shared library cache 46.33 37.60 -18.84
shared sql area 340.77 -100.00
streams Sender info 2.82 2.82 0.00
streams free memory 15.67 15.67 0.00
streams kwqbcqini:spilledovermsgs 0.45 0.45 0.00
streams kwqbdaspl:spilledovermsgs 1.69 1.69 0.00
buffer_cache 17,264.00 17,712.00 2.59
fixed_sga 2.02 2.02 0.00
log_buffer 13.98 13.98 0.00
确定下是否存在内存抖动 share pool抖动明显
页:
[1]