legato networker 备份oracle RAC

daniel_w

7.2.1的,NMO 4.2,HP-UX 11.23,都是在兼容列表内的.也不知道是什么原因,就是最后的index,bootstrap就没有动作了,一直是运行状态,也不报错,也不运行.

yddll

是Lanfree，一个驱动器吧\r\n单独建一个pool来备份索引呢？\n\n[ 本帖最后由 yddll 于 2006-3-30 21:38 编辑 ]

daniel_w

现在还没有做到LANFREE那一步呢,HP ESL712 4个驱动器.是啊,是用的Default存放index,bootstrap的,在这个环境中也有几个ORACLE,文件系统的备份都是没有问题的.而且在出现这个问题的机器上备份文件都是没有问题的,写完备份的文件就会保存index,bootstrap到Default 池.实质上这个RMAN的备份是完成的,通过/nsr/applogs/nsrnmo.log是可以看出的,在monitor也可以看出RMAN的写操作已完成,但是就是在要保存index,bootstrap时就不做任何操作了,也不报错,无错查啊,这真是不留一点痕迹啊.我手工将该备份组STOP后,它会对这个STOP的备份组做savegrp -O的操作,因为RMAN备份完成,所以它会把RMAN写入INDEX,并且在INDEX中会显示出RMAN的备份内容,使利用RMAN恢复数据还可以正常的恢复.\r\n\r\n  我现在做了一个savegrp -D 9 -O \"rac group\"看它的DEBUG过程呢,RMAN备份是完成了,但是在写INDEX,BOOTSTRAP时像似有个错误RPC:is not checksum,正在查找错误原因呢,怪事,我通过在节点手工执行RMAN脚本,然后在NETWORKER执行savegrp -O \"rac group\"都没问题,而且备份文件也可以自动保存index,bootstrap.

daniel_w

你说的方法我都已经试过了.我差不多什么办法都试了.没有效果.哪怕是个错误提示,也有证可依啊.\r\n我现在用savegrp -D 9 -O \"rac group\"来分析呢,正在看在哪里出现的问题呢.继续努力中......

daniel_w

这个是daemon.log关于这次备份的过程。证明RMAN脚本已执行成功。接下来的动作index,bootstrap就没了。\r\n\r\n14:45:44 nsrd: savegroup info: starting  RacFull (with 1 client(s))\r\n03/31/06 14:49:43 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh saving to pool \'RacPool\' (HQ3931L3)\r\n03/31/06 14:55:10 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh done saving to pool \'RacPool\' (HQ3931L3) 4876 MB\r\n03/31/06 14:55:19 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh saving to pool \'RacPool\' (HQ3931L3)\r\n03/31/06 14:55:20 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh done saving to pool \'RacPool\' (HQ3931L3) 9728 KB\r\n03/31/06 14:55:29 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh saving to pool \'RacPool\' (HQ3931L3)\r\n03/31/06 14:55:29 nsrd: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh done saving to pool \'RacPool\' (HQ3931L3) 11 MB\r\n03/31/06 14:56:08 nsrd: write completion notice: Writing to volume HQ3931L3 complete

daniel_w

这个是nsrnmostart.log的日志，也可以看出RMAN以正常运行。\r\n\r\nargc = 16\r\nCalling Summary\r\nargv[ 0] = nsrnmostart\r\nargv[ 1] = -s\r\nargv[ 2] = backupsrv\r\nargv[ 3] = -g\r\nargv[ 4] = RacFull\r\nargv[ 5] = -LL\r\nargv[ 6] = -m\r\nargv[ 7] = rx8620a\r\nargv[ 8] = -l\r\nargv[ 9] = full\r\nargv[10] = -q\r\nargv[11] = -W\r\nargv[12] = 78\r\nargv[13] = -N\r\nargv[14] = RMAN:/usr/bin/or10g_full_bk.sh\r\nargv[15] = RMAN:/usr/bin/or10g_full_bk.sh\r\nEnvironment Read by nsrnmostart\r\nORACLE_HOME          = /oracle/product/10g\r\nPRECMD               = \r\nPOSTCMD              = \r\nPATH                 = /opt/networker/bin:/oracle/product/10g/bin:/usr/bin:/bin:/usr/sbin\r\nNSR_RMAN_ARGUMENTS   = \r\nNSR_RMAN_OUTPUT      = \r\nnwora_parse_msglog_name:  is not in a valid format.\r\nLeaving Function nwora_process_calling_args\r\n\r\nEntering Function nwora_scan_rman_script\r\nThe RMAN script is: /usr/bin/or10g_full_bk.sh\r\nChecking rman script /usr/bin/or10g_full_bk.sh for validity.\r\nfound connect target string.\r\nfound connect catalog string.\r\nfound allocate channel: {allocate channel ch00 type \'sbt_tape\';\r\nCompleted checking of rman script.\r\nLeaving Function nwora_scan_rman_script\r\n\r\nPre-processing command succeeded.\r\n\r\nPre-processing command succeeded.\r\n\r\nEntering Function nwora_nsrnmostart_rman\r\nnwora_set_RMAN_location: Setting the location of RMAN to \'/oracle/product/10g/bin\'.\r\nnwora_set_RMAN_location: The location of RMAN is: /oracle/product/10g/bin\r\nnwora_set_RMAN_location: The name of the RMAN binary is: rman\r\nnwora_set_RMAN_args: Setting the RMAN args:\r\n        send \'NSR_ENV=(NSR_SERVER=backupsrv,\r\nNSR_GROUP=RacFull,\r\nNSR_SAVESET_NAME=RMAN:/usr/bin/or10g_full_bk.sh)\'\r\nnwora_set_RMAN_args: 1 arguments set.\r\ncmd_args = send \'NSR_ENV=(NSR_SERVER=backupsrv,\r\nNSR_GROUP=RacFull,\r\nNSR_SAVESET_NAME=RMAN:/usr/bin/or10g_full_bk.sh)\'\r\nrman_script = /usr/bin/or10g_full_bk.sh\r\nsaveset_name = RMAN:/usr/bin/or10g_full_bk.sh\r\nLaunching backup process\r\nnwora_run_RMAN: Writing out the RMAN script.\r\nnwora_run_RMAN: The argv[] is:\r\n        /oracle/product/10g/bin/rman\r\n        send \'NSR_ENV=(NSR_SERVER=backupsrv,\r\nNSR_GROUP=RacFull,\r\nNSR_SAVESET_NAME=RMAN:/usr/bin/or10g_full_bk.sh)\'\r\n        cmdfile\r\n        \'/usr/bin/or10g_full_bk.sh\'\r\nnwora_run_RMAN: The envp is:\r\nnwora_run_RMAN: Spawning the RMAN session.\r\nnwora_spawn_RMAN: Creating the communication pipe.\r\nnwora_spawn_RMAN: Making the pipe non-blocking.\r\nnwora_spawn_RMAN: Spawning the RMAN process.\r\nnwora_spawn_RMAN: Spawned the RMAN process 23335.\r\nrman returned success\r\nLeaving Function nwora_nsrnmostart_rman\r\n\r\nPost-processing command succeeded.

daniel_w

刚才突然冒出个想法，nsrnmo真的就正常退出了吗？是不能保存index and bootstrap还是nsrnmo没有正常退出？做了个验证。在节点上执行：ps -ef |grep -i nsr  看到了正在运行的进程,用kill -9 killed。大家有没有看到什么？killed后可以做index,bootstrap的保存。那就是说问题不在保存index,bootstrap上，而是在nsrnmo 调用 nsrnmostart没有正常的发送成功退出通知。\r\n\r\n15:36:37 nsrd: savegroup alert: Group RacFull aborted, savegroup is already running\r\n03/31/06 15:46:53 savegrp: command \'nsrnmo -s backupsrv -g RacFull -LL -m rx8620a -l full -q -W 78 -N RMAN:/usr/bin/or10g_full_bk.sh RMAN:/usr/bin/or10g_full_bk.sh \' for client rx8620a exited with return code 1.\r\n03/31/06 15:46:53 savegrp: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh will retry 1 more time(s)\r\n03/31/06 15:47:35 savegrp: command \'nsrnmo -s backupsrv -g RacFull -LL -m rx8620a -l full -q -W 78 -N RMAN:/usr/bin/or10g_full_bk.sh RMAN:/usr/bin/or10g_full_bk.sh \' for client rx8620a exited with return code 1.\r\n03/31/06 15:47:35 savegrp: rx8620a:RMAN:/usr/bin/or10g_full_bk.sh will retry 0 more time(s)\r\n03/31/06 15:47:36 nsrd: backupsrv:index:rx8620a saving to pool \'Index\' (HQ3944L3)\r\n03/31/06 15:47:36 nsrd: backupsrv:index:rx8620a done saving to pool \'Index\' (HQ3944L3) 20 KB\r\n03/31/06 15:47:37 nsrd: backupsrv:bootstrap saving to pool \'Index\' (HQ3944L3)\r\n03/31/06 15:47:37 nsrmmdbd: media db is saving its data.  This may take a while.\r\n03/31/06 15:47:37 nsrmmdbd: media db is open for business.\r\n03/31/06 15:47:38 nsrd: backupsrv:bootstrap done saving to pool \'Index\' (HQ3944L3) 285 KB\r\n03/31/06 15:47:38 nsrd: savegroup info: Added \'backupsrv\' to the group \'RacFull\' for bootstrap backup.\r\n03/31/06 15:47:38 nsrd: savegroup alert: RacFull completed, total 2 client(s), 0 Hostname(s) Unresolved, 1 Failed, 1 Succeeded. (rx8620a Failed)\r\n* rx8620a:RMAN:/usr/bin/or10g_full_bk.sh /opt/networker/bin/nsrnmo[272]: 23334 Killed

旋子

那是不是找出问题出在哪里了?

daniel_w

应该是吧。经过分析测试基本上可以定位到这一点上。我有个预感，不会是NMO 4.2不支持\r\nORACLE 10G 10.1.0.2.0啊。问题这么怪。\n\n[ 本帖最后由 daniel_w 于 2006-3-31 16:50 编辑 ]

旋子

呵呵,但愿你的预感是正确的!不管怎样,是有点头绪了嘛!