Oracle数据库4031故障分析

来源: https://www.ucloud.cn/yun/129090.html 作者: 不知名网友 发布日期: 发布于2023-01-11 11:02

Oracle数据库4031故障分析

点击上方“IT那活儿**”公众号，关注后了解更多内容，不管IT什么活儿，干就完了！！！**

一

问题描述

数据库业务出现超时，数据库出现enq: TX - index contention等等待事件。

二

问题分析

2.1 数据库出现enq: TX - index contention等待

通过查看问题前的ash，发现在数据库异常之前，数据库的主要等待事件是：SGA: allocation forcing component growth。

Alert.log：

2022-10-03T00:16:41.029338+08:00Errors in file /u01/app/oracle/diag/rdbms/*/trace/db2_ora_6519.trcORA-04031: unable to allocate 63496 bytes of shared memory ("shared pool","DBMS_BACKUP_RESTORE","PLMCD^a5f238a0","BAMIMA: Bam Buffer")Incident details in: /u01/app/oracle/diag/rdbms/*/incident/incdir_218692/db2_ora_6519_i218692.trcUse ADRCI or Support Workbench to package the incident.See Note 411.1 at My Oracle Support for error and packaging details.2022-10-03T00:16:41.321084+08:00ORA-04031 heap dump being written to trace file /u01/app/oracle/diag/rdbms/*/incident/incdir_218692/db2_ora_6519_i218692.trc2022-10-03T00:19:38.948488+08:00Errors in file /u01/app/oracle/diag/rdbms/*/trace/b2_m000_8030.trc (incident=218693):ORA-04031: unable to allocate 57768 bytes of shared memory ("shared pool","DBMS_STATS_INTERNAL","PLMCD^60dfd26c","BAMIMA: Bam Buffer")Incident details in: /u01/app/oracle/diag/rdbms/*/incident/incdir_218693/b2_m000_8030_i218693.trcUse ADRCI or Support Workbench to package the incident.See Note 411.1 at My Oracle Support for error and packaging details.2022-10-03T00:19:42.233916+08:00Errors in file /u01/app/oracle/diag/rdbms/*/trace/db2_ora_8166.trc (incident=218700):ORA-04031: unable to allocate 63496 bytes of shared memory ("shared pool","DBMS_BACKUP_RESTORE","PLMCD^a5f238a0","BAMIMA: Bam Buffer")Incident details in: /u01/app/oracle/diag/rdbms/*/incident/incdir_218700/db2_ora_8166_i218700.trcUse ADRCI or Support Workbench to package the incident.See Note 411.1 at My Oracle Support for error and packaging details.…2022-10-03T04:23:42.937853+08:00Errors in file /u01/app/oracle/diag/rdbms/*/trace/db2_lmd0_7330.trc (incident=215852):ORA-04031: unable to allocate 13840 bytes of shared memory ("shared pool","unknown object","sga heap(1,0)","ges resource dynamic")Use ADRCI or Support Workbench to package the incident.See Note 411.1 at My Oracle Support for error and packaging details.2022-10-03T04:23:49.345977+08:00Errors in file /u01/app/oracle/diag/rdbms/*/trace/db2_lmd0_7330.trc (incident=215853):ORA-04031: unable to allocate 13840 bytes of shared memory ("shared pool","unknown object","sga heap(1,0)","ges resource dynamic")Use ADRCI or Support Workbench to package the incident.See Note 411.1 at My Oracle Support for error and packaging details.…2022-10-03T08:58:07.793554+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T09:08:09.957747+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T09:18:12.005452+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T09:28:12.835118+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)022-10-03T09:38:13.206769+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T09:44:00.615729+08:00Thread 2 advanced to log sequence 29428 (LGWR switch)Current log# 8 seq# 29428 mem# 0: +DG_DATA_SSD_1/N*/ONLINELOG/group_8.276.1008547021Current log# 8 seq# 29428 mem# 1: +DG_DATA_SSD_1/N*/ONLINELOG/group_8.277.10085470252022-10-03T09:44:01.334950+08:00TT03: Standby redo logfile selected for thread 2 sequence 29428 for destination LOG_ARCHIVE_DEST_22022-10-03T09:44:03.799630+08:00Archived Log entry 119834 added for T-2.S-29427 ID 0x79a6de54 LAD:12022-10-03T09:48:17.418439+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T09:58:19.513479+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T10:08:21.022510+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T10:14:50.702688+08:00LCK1 (ospid: 7342) waits for event libcache interrupt action by LCK for 73 secs.2022-10-03T10:14:50.702867+08:00LCK1 (ospid: 7342) is hung in an acceptable location (libcache 0x41.02).2022-10-03T10:17:11.454316+08:00LCK1 (ospid: 7342) waits for event libcache interrupt action by LCK for 214 secs.2022-10-03T10:17:11.454494+08:00LCK1 (ospid: 7342) is hung in an acceptable location (libcache 0x41.02).2022-10-03T10:18:23.645094+08:00DDE: Problem Key ORA 4031 was completely flood controlled (0x6)Further messages for this problem key will be suppressed for up to 10 minutes2022-10-03T10:21:42.845553+08:00LCK1 (ospid: 7342) waits for event libcache interrupt action by LCK for 79 secs.2022-10-03T10:21:42.845731+08:00LCK1 (ospid: 7342) is hung in an acceptable location (libcache 0x41.02).2022-10-03T10:24:20.692124+08:00LCK1 (ospid: 7342) waits for event libcache interrupt action by LCK for 80 secs.2022-10-03T10:24:20.692681+08:00LCK1 (ospid: 7342) is hung in an acceptable location (libcache 0x41.02).2022-10-03T10:27:57.458418+08:00CKPT (ospid: 7350) waits for event enq: XR - database force logging for 78 secs.2022-10-03T10:28:07.628285+08:00CKPT (ospid: 7350) waits for event enq: XR - database force logging for 88 secs.2022-10-03T10:28:17.668429+08:00CKPT (ospid: 7350) waits for event enq: XR - database force logging for 98 secs.

2.2 查看数据库内存使用情况

通过ash趋势和alert.log发现，数据库在出现enq: TX - index contention之前，数据库主要等待是SGA: allocation forcing component growth和library cache lock，并且alert.log从0点开始出现ora-4031持续到异常时段。

Buffer cache的内存非常小，只有128m，而shared pool占用了32G。

Shared pool中ges resource dynamic和ges enqueues分别占用了12G和8.9G，这通常是不正常的。说明由于这两个组件占用了过多的内存，导致数据库的ORA-4031，进而出来sga resize，进而影响了数据库的整体性能。

分析lmhb.trc：

grep "library cache pin wait check" n*2_lmhb_7340.trckjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 failedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 failedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 failedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 failedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 failedkjgcr_ServiceGCR: KJGCR_METRICS: Local metric library cache pin wait check, id 11 succeed…grep "check lck heartbeat" n*2_lmhb_7340.trckjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_ChkGlobalMetric: metric 7 (check lck heartbeat) failed for dbname N*, inst 2, node 2kjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growthkjgcr_GrowResourceCache: LCP waits high, allowing res cache growth

参考文档：Error ORA-04031 in the Shared Pool with Huge Allocation in Memory Type "ges resource dynamic" or "ges enqueues" memory" (Doc ID 2631592.1)

For Oracle Versions >= 12.2 but BELOW 19.1Bug:26405036 - VERY HIGH "GES ENQUEUES" ON THE SHARED POOL

认为命中Bug:26405036 - VERY HIGH "GES ENQUEUES" ON THE SHARED POOL和Bug:27824540 - ORA-04031 ("SHARED POOL","UNKNOWN OBJECT","SGA HEAP(1,0)","GES RESOURCE DYNAMIC"。

其中Bug:26405036 fix在12.2.0.1.200114  ，Bug:27824540 fix在12.2.0.1.190716。

2.3 对应的解决方案

For Oracle Versions >= 12.2 but BELOW 19.1Apply fix for Bug:26405036; see NOTE:26405036.8Workaround: on R12.2 or above, Start pseudo reconfiguration by below command is workaround.SQL> oradebug setmypidSQL> oradebug lkdebug -m reconfig lkdebugApply fix for Bug:27824540; see NOTE:27824540.8Workaround: There are 2 possible workarounds.1) Disable the action 11.SQL> oradebug dyn_gcr -a 11 -disableNote: This oradebug command is available on 12.2 and later.2) Disable the GES resource cache; set the initialization parameter "_ges_direct_free" to TRUE.Note: Completely disabling GES resource cache may lead to some other side affects like contention on TM lock for insert statements,so use this workaround with caution if there is still a need.

三

问题总结

数据库由于shared pool中ges resource dynamic和ges enqueues占用内存过多，shared pool不够用引起ora-4031和sga resize，进而影响了数据库整体性能。

通过重启实例后，ges resource dynamic空间释放，但ges enqueues还有8g，目前看已经缓解问题。最终建议通过补丁解决。

建 议：

oracle 12.2已过支持期，但基于hp-ux平台oracle 12.2.0.1.180417的26405036、27824540的merge patch不存在并且可能无法申请出来。

• 建议1

  只安装补丁26405036 - VERY HIGH "GES ENQUEUES" ON THE SHARED POOL看问题是否解决。

• 建议2

  升级RU到12.2.0.1.200114以上版本，已修复以上问题。

• 建议3

  尝试使用bug中的workaround或重启实例做为临时的解决方案：

SQL> oradebug setmypidSQL> oradebug lkdebug -m reconfig lkdebug

优先建议使用建议1，小补丁窗口时间短并且可以rolling。

本文作者：张 芳(上海新炬中北团队）

本文来源：“IT那活儿”公众号