1. 首先,确定你的硬件错误属于下面的一些问题.
T 可以使用dd命令来读一些物理设备上的数据.比如:读第一个100MB从c1t11d0设备上.
# dd if=/dev/rdsk/c1t11d0s2 of=/dev/null bs=1024k count=100
.一些磁盘阵列要求你的硬件的cache设置为disable.请先设置这个参数.
2. 当物理磁盘损坏,分配在这个磁盘上的子磁盘会detached.使用vxprint –ht命令显示子磁盘如"NDEV" (No Device)和RAID-5卷的操作为"Degraded Mode".VXVM的log信息会出现在/var/adm/messages中.比如:
Nov 9 16:05:29 sydvcs4 unix: WARNING: vxvm:vxio: object disk1-01 detached from RAID-5 r5vol at column 0 offset 0
Nov 9 16:05:29 sydvcs4 unix: WARNING: vxvm:vxio: RAID-5 r5vol entering degraded mode operation
对于ANY WRITE OPERATION来说这个理解是非常重要的,在卷进入"Degraded Mode"的时候,那些被detached的子磁盘上的数据.当物理磁盘被重新加入到
系统中的时候写的数据被
存储到RAID-5奇偶校验上的.这些数据需要从奇偶校验上恢复到新的子磁盘上.
3. 当detached的子磁盘或者奇偶校验上的数据copy是可以用的,允许设置一个预先的奇偶校验.音问只有当卷是disable的时候奇偶校验才是disable的,使用一个最新的用户数据.一个通常的RAID-5上恢复数据的方法是重新计算detach子磁盘上的奇偶校验,在故障数据上重新写可用数据. This kind of corruption is IRREVERSIBLE
4. RAID-5卷可以通过delayrecover参数来开始.卷管理可以推迟恢复直到执行vxrecover命令. (The recovery can either be parity recalculation or regenerating the data from parity, both have the potential to overwrite valid data.)
# vxvol -g
-o delayrecover start
可以使用 vxstat 命令来查看子磁盘的写操作:
# vxstat -i 5 -g -s ....
通过运行T vxprint -ht -g . 来获得子磁盘的信息 列出子磁盘信息可以如下命令:
# vxprint -ht -g | awk '/^sd/ {print $2}
/^s2/ {print $2}'
5. 不用强行进行RAID5卷的开始的情况是应为卷管理已经预先设置的detached子磁盘(DET, FAIL or RCOV状态)开始奇偶校验会重新计算重新写奇偶校验.如果在一行中出现了2个或者更多的detached子磁盘.卷管理并不知道这些子磁盘包含up-to-date数据.如果,你没有加delayrecover参数的话, 奇偶校验重新计算并不会开始.同时卷会进入NEEDSYNC状态.如果强行进行卷启动并且没有加delayrecover参数的话, 奇偶校验重新计算会立刻开始.如果奇偶校验包括更多的up-to-date用户数据,着卷将损坏.
6. RAID5的log是在奇偶校验和用户数据之间连续性的一个保证.RAID5的log永远是replay的.如果要求进行奇偶校验确认(vxr5check)则连续性是不能保证的.
7. The VERITAS File System (VxFS) log is crucial for the recovery of an interrupted VxFS file system. Without replaying the log, file system corruption will happen, especially if the file system is heavily used when the RAID-5 volume goes down. The catch is that the VxFS log needs to be replayed and WRITTEN to the RAID-5 volume before the full file system check can proceed, but without knowing the integrity of the RAID-5 volume, any write operation may cause corruption to volume data. In order to solve the dilemma, metasave can be used to get an image of the file system and run fsck again the saved image before running the actual fsck on the RAID-5 volume. The metasave utility can be obtained from the VERITAS ftp site: metasave is OS version dependent, please choose the right version for your OS. For example, when you run it on Solaris 8, the following command can get a copy of the file system structure to an ordinary file and a fsck can run on the ordinary file. Remember to replay the log, DO NOT use the "nolog" option.
# metasave_5.8 /dev/vx/rdsk// | metasave_5.8 -r
# fsck -F vxfs -o full
---------------------------------------------------------------------------------
RAID-5 卷中的任何一行, 如果子磁盘的状态为:"NDEV", "FAIL", "DET" or "RCOV", 则子磁盘的数据可以从奇偶校验中恢复. 比如:RAID-5 卷的状态如下:
v r5vol - ENABLED ACTIVE 24576000 raid5 - RAID
pl r5vol-01 r5vol ENABLED ACTIVE 24577920 RAID 5/32 RW
sd disk1-01 r5vol-01 disk1 0 2049120 0/0 - NDEV <<<< 1st ROW with OFFSET 0
sd disk10-01 r5vol-01 disk10 415 2047680 0/2049120 sdc1t1d0s0 ENA <<<< 2nd ROW with OFFSET 2049120
sd disk15-01 r5vol-01 disk15 415 2047680 0/4096800 sdc1t1d0s5 ENA <<<< 3rd ROW with OFFSET 4096800
sd disk3-01 r5vol-01 disk3 0 2049120 1/0 c1t3d0 ENA <<<< 1st ROW with OFFSET 0
sd disk30-01 r5vol-01 disk30 415 2047680 1/2049120 - NDEV <<<< 2nd ROW with OFFSET 2049120
sd disk35-01 r5vol-01 disk35 415 2047680 1/4096800 sdc1t3d0s5 ENA <<<< 3rd ROW with OFFSET 4096800
sd disk4-01 r5vol-01 disk4 0 2049120 2/0 c1t4d0 ENA <<<< 1st ROW
sd disk40-01 r5vol-01 disk40 415 2047680 2/2049120 sdc1t4d0s0 ENA
sd disk45-01 r5vol-01 disk45 415 2047680 2/4096800 - NDEV
sd disk5-01 r5vol-01 disk5 0 2049120 3/0 c1t5d0 ENA <<<< 1st ROW
sd disk50-01 r5vol-01 disk50 415 2047680 3/2049120 sdc1t5d0s0 ENA
sd disk55-01 r5vol-01 disk55 415 2047680 3/4096800 sdc1t5d0s5 ENA
sd disk6-01 r5vol-01 disk6 0 2049120 4/0 c1t6d0 ENA <<<< 1st ROW
sd disk60-01 r5vol-01 disk60 415 2047680 4/2049120 sdc1t6d0s0 ENA
sd disk65-01 r5vol-01 disk65 415 2047680 4/4096800 sdc1t6d0s5 ENA
pl r5vol-02 r5vol ENABLED LOG 5760 CONCAT - RW
sd disk56-01 r5vol-02 disk56 415 5760 0 sdc1t5d0s6 ENA
第一行的 disk1-01, disk3-01, disk4-01, disk5-01和disk6-01在disk1-01 上只有一个子磁盘是failed. disk30-01在第2行只有一个子磁盘是failed disk45-01 在第3行上只有一个子磁盘是failed. 恢复这个过程只需要简单的reaattach或者替换坏的磁盘,然后,运行vxrecover.
如果卷是停止的则需要start.使用delayrecover参数启动卷.
# vxvol -g -o delayrecover
The end-user should check the data integrity without writing any data to the volume. In case of VERITAS File System (VxFS), because the intent log need to be replayed, write operation is required. If you want to be completely safe, get the VxFS image with metasave and run fsck against the image.
If you are confident that the volume data is in good shape, just run a simple fsck on the file system:
# fsck -F vxfs /dev/vx/rdsk//
If a full file system check is required even AFTER THE LOG IS REPLAYED, run a full fsck with option "-n" to assess the severity of damage:
# fsck -F vxfs -o full -n /dev/vx/rdsk//
On the other hand, if want to be sure that the volume is recovered properly before replaying the VxFS intent log, you can save an image of the file system to an ordinary file and run a full fsck on that ordinary file. Please note that a log replay is essential in the recovery of VxFS, the nolog option should not be used unless the intent log is corrupted.
# metasave_5.8 /dev/vx/rdsk// | metasave_5.8 -r
# fsck -F vxfs -o full
If the VxFS image is checked okay, then you can run the fsck on the actual volume.
当校验完数据,通过vxreattach来恢复了detached的子磁盘,或者使用vxdiskadm来替换故障磁盘.如果子磁盘是"RCOV"状态.vxrecover能够恢复子磁盘.
恢复在一行中出现2个或者以上故障子磁盘.
------------------------------------------------------------------------------------------------------------------------------------------------
比如::
v r5vol - DETACHED ACTIVE 24576000 raid5 - RAID
pl r5vol-01 r5vol ENABLED ACTIVE 24577920 RAID 5/32 RW
sd disk1-01 r5vol-01 disk1 0 2049120 0/0 - NDEV
sd disk10-01 r5vol-01 disk10 415 2047680 0/2049120 sdc1t1d0s0 ENA
sd disk15-01 r5vol-01 disk15 415 2047680 0/4096800 sdc1t1d0s5 ENA
sd disk3-01 r5vol-01 disk3 0 2049120 1/0 c1t3d0 ENA
sd disk30-01 r5vol-01 disk30 415 2047680 1/2049120 - NDEV
sd disk35-01 r5vol-01 disk35 415 2047680 1/4096800 sdc1t3d0s5 ENA
sd disk4-01 r5vol-01 disk4 0 2049120 2/0 c1t4d0 ENA
sd disk40-01 r5vol-01 disk40 415 2047680 2/2049120 sdc1t4d0s0 ENA
sd disk45-01 r5vol-01 disk45 415 2047680 2/4096800 - NDEV
sd disk5-01 r5vol-01 disk5 0 2049120 3/0 c1t5d0 ENA
sd disk50-01 r5vol-01 disk50 415 2047680 3/2049120 sdc1t5d0s0 ENA
sd disk55-01 r5vol-01 disk55 415 2047680 3/4096800 sdc1t5d0s5 ENA
sd disk6-01 r5vol-01 disk6 0 2049120 4/0 - NDEV
sd disk60-01 r5vol-01 disk60 415 2047680 4/2049120 - NDEV
sd disk65-01 r5vol-01 disk65 415 2047680 4/4096800 - NDEV
pl r5vol-02 r5vol ENABLED LOG 5760 CONCAT - RW
sd disk56-01 r5vol-02 disk56 415 5760 0 sdc1t5d0s6 ENA
在 /var/adm/messages 文件中有以下信息:, t
Nov 9 16:05:29 sydvcs4 unix: WARNING: vxvm:vxio: object disk1-01 detached from RAID-5 r5vol at column 0 offset 0
Nov 9 16:05:29 sydvcs4 unix: WARNING: vxvm:vxio: RAID-5 r5vol entering degraded mode operation
Nov 9 16:07:30 sydvcs4 unix: WARNING: vxvm:vxio: object disk30-01 detached from RAID-5 r5vol at column 1 offset 2049120
Nov 9 16:07:30 sydvcs4 unix: WARNING: vxvm:vxio: RAID-5 r5vol entering degraded mode operation
Nov 9 16:07:30 sydvcs4 unix: WARNING: vxvm:vxio: object disk45-01 detached from RAID-5 r5vol at column 1 offset 2049120
Nov 9 16:07:30 sydvcs4 unix: WARNING: vxvm:vxio: RAID-5 r5vol entering degraded mode operation
(24 minutes later)
Nov 9 16:31:29 sydvcs4 unix: WARNING: vxvm:vxio: Double failure condition detected on RAID-5 r5vol
Nov 9 16:31:29 sydvcs4 unix: WARNING: vxvm:vxio: detaching RAID-5 r5vol
Another place to find out the failure sequence is in those emails sent by the vxnotify process. By default the mails go to root email box. Look for the following emails. For example, the following messages were sent to root because of the disk failures:
-------- Begin of root mails ----------
Message 63:
From root Sat Nov 9 16:05:45 2002
Date: Sat, 9 Nov 2002 16:05:45 +1100 (EST)
From: Super-User
To: root
Subject: Volume Manager failures on host sydvcs4
Failures have been detected by the VERITAS Volume Manager:
failed disks:
disk1
failing disks:
disk1
Message 65:
From root Sat Nov 9 16:07:47 2002
Date: Sat, 9 Nov 2002 16:07:47 +1100 (EST)
From: Super-User
To: root
Subject: Volume Manager failures on host sydvcs4
Failures have been detected by the VERITAS Volume Manager:
failed disks:
disk1
disk30
failing disks:
disk1
disk30
Message 68:
From root Sat Nov 9 16:13:27 2002
Date: Sat, 9 Nov 2002 16:13:27 +1100 (EST)
From: Super-User
To: root
Subject: Volume Manager failures on host sydvcs4
Failures have been detected by the VERITAS Volume Manager:
failed disks:
disk1
disk30
disk45
failing disks:
disk1
disk30
(24 minutes later)
Message 4:
From root Sat Nov 9 16:37:04 2002
Date: Sat, 9 Nov 2002 16:37:04 +1100 (EST)
From: Super-User
To: root
Subject: Volume Manager failures on host sydvcs4
Failures have been detected by the VERITAS Volume Manager:
failed disks:
disk1
disk30
disk45
disk60
disk65
failing disks:
disk1
disk30
.......
---------- End of root emails --------
From the above emails, it can be seen that disk1, disk30 and disk45 failed before disk6, disk60 and disk65.
According to the /var/adm/messages logs and vxnotify emails, it is apparent that there were 24 minutes lapsed from the point the volume entered "degraded mode" to the point the volume was disabled (detached) due to double failure. It can be safely assumed that subdisks disk1-01, disk10-01 and disk45-01 were obsoleted by the write operations which occurred during those 24 minutes. The recovery procedures should be like the following.
1. 保存磁盘组的配置:
# vxdisk -o alldgs list
# vxprint -rht
# vxprint -Am
2. 分离故障子磁盘确保你备份了原始的配置,以便恢复时使用)
# vxsd -g dis .....
列如从 /var/adm/messages文件中来的信息.
# vxsd -g dis disk1-01 disk30-01 disk45-01
3. Unstale子磁盘的可用数据:
# vxmend -g unstale ...
以上列来看:
# vxmend -g fix unstale disk6-01 disk60-01 disk65-01
4. 通过delayrecover参数启动卷:
# vxvol -g -o delayrecover start
5. 要求用户进行数据检验
6. If the data is verified by the end-user, the obsolete subdisks can be associated back. You will need the original configuration (vxprint -ht) in order to obtain the correct position to put the subdisks back. The / values can be found in the 7th column of the vxprint -ht output.
# vxsd -g -l / assoc
For the above example, associate back disk1-01, disk30-01 and disk45-01 according to the saved vxprint -ht output.
# vxsd -g -l 0/0 assoc r5vol-01 disk1-01
# vxsd -g -l 1/2049120 r5vol-01 disk30-01
# vxsd -g -l 2/4096800 r5vol-01 disk45-01
The association of a subdisk should automatically start the recovery of data to the subdisks from the parity. Check the tasks with the vxtask list command.
Recovery with two or more failed subdisks in each row and the sequence of failure can not be determined
--------------------------------------------------------------------------------------------------------------------------------------------------
If for whatever reason the sequence of failure of subdisks can not be determined, the RAID-5 needs to be recovered by trial and error. Please keep a detailed log on what you have done because it will be a very tedious task.
It is important to remember that once a subdisk got overwritten, the data is gone forever. The same is true for the parity, once the parity is recalculated, the data stored there during degraded mode is gone forever. Never force the volume to start with "-f" or -o force" option and always start the volume in the delayrecover mode.
Always be careful not to enable (unstale) all the subdisks in the same row. If all the subdisks are enabled manually on the same row, the parity may become inconsistent with the data, this will lead to data corruption later on.
It is NOT correct just to fix the plex to empty by running vxmend and starting the volume, because then the parity will be recalculated and whatever data stored in the parity during degraded mode will be wiped out.
The procedure to recover the volume in this kind of situation is listed below.
1. Save a copy of the disk group configuration:
# vxdisk -o alldgs list
# vxprint -rht
# vxprint -Am
2. Try to find out as much as possible from the /var/adm/messages and vxnotify email on what subdisks are obsolete, then dissociate them from the plex.
# vxsd -g dis
Take the following volume as an example. If it is known that disk1-01 is obsolete, disk1-01 should be dissociated. Make sure you have the vxprint -ht output before detaching the subdisk, because later the subdisk has to be put back to the original location.
# vxsd -g dis disk1-01
and unstale the disk6-01 which is in the same row:
# vxmend -g fix unstale disk6-01
v r5vol - DETACHED ACTIVE 24576000 raid5 - RAID
pl r5vol-01 r5vol ENABLED ACTIVE 24577920 RAID 5/32 RW
sd disk1-01 r5vol-01 disk1 0 2049120 0/0 - NDEV <<<< first row
sd disk10-01 r5vol-01 disk10 415 2047680 0/2049120 sdc1t1d0s0 ENA
sd disk15-01 r5vol-01 disk15 415 2047680 0/4096800 sdc1t1d0s5 ENA
sd disk3-01 r5vol-01 disk3 0 2049120 1/0 c1t3d0 ENA
sd disk30-01 r5vol-01 disk30 415 2047680 1/2049120 - NDEV <<< second row
sd disk35-01 r5vol-01 disk35 415 2047680 1/4096800 sdc1t3d0s5 ENA
sd disk4-01 r5vol-01 disk4 0 2049120 2/0 c1t4d0 ENA
sd disk40-01 r5vol-01 disk40 415 2047680 2/2049120 sdc1t4d0s0 ENA
sd disk45-01 r5vol-01 disk45 415 2047680 2/4096800 - NDEV <<< third row
sd disk5-01 r5vol-01 disk5 0 2049120 3/0 c1t5d0 ENA
sd disk50-01 r5vol-01 disk50 415 2047680 3/2049120 sdc1t5d0s0 ENA
sd disk55-01 r5vol-01 disk55 415 2047680 3/4096800 sdc1t5d0s5 NDEV <<< third row
sd disk6-01 r5vol-01 disk6 0 2049120 4/0 - NDEV <<< first row
sd disk60-01 r5vol-01 disk60 415 2047680 4/2049120 - NDEV <<< second row
sd disk65-01 r5vol-01 disk65 415 2047680 4/4096800 - NDEV <<< third row
pl r5vol-02 r5vol ENABLED LOG 5760 CONCAT - RW
sd disk56-01 r5vol-02 disk56 415 5760 0 sdc1t5d0s6 ENA
3. For the remaining two rows, dissociate one subdisk from each row at a time, start the volume with the delayrecover option and verify the data. This has to be done as many time as required until all the valid subdisks can be located.
If there are N failed subdisks in first row, M failed subdisks in second row and P failed subdisks in third row, it may be necessary to try M x N x P times before all the valid data can be found.
For the above example, after taking out disk1-01 from the first row, it is clear that disk6-01 has the valid data. For the second row, disk30-01 and disk60-01 should be taken out, and for the third row, disk45-01, disk55-01 and disk65-01. Disks in all the following 6 combinations should be taken out.
disk30-01 disk45-01
disk30-01 disk55-01
disk30-01 disk65-01
disk60-01 disk45-01
disk60-01 disk55-01
disk60-01 disk65-01
For the first attempt, dissociate disk30-01 and disk45-01:
# vxsd -g dis disk30-01 disk45-01
Then enable (unstale) the other subdisks and start the volume with the delayrecover option:
# vxmend -g fix unstale disk60-01 disk55-01 disk65-01
# vxvol -o delayrecover -g start
Then verify data using methods that will not write any data to the volume. In the case of VxFS, use metasave to take an image and run fsck on the image.
# metasave_5.8 /dev/vx/rdsk// | metasave_5.8 -r
# fsck -F vxfs -o full
If fsck returns without error, then fsck can be run on the actual volume.
If fsck returns lots of errors, it means there are obsolete subdisks, the next set of subdisks needs to be taken out, the previous dissociated subdisks need to be put back. It is safer to dissociate the next set of subdisks before associating the original set of subdisks because, if the volume is started and a subdisk is associated, Volume Manager will start the recovery right away and overwrite the newly associated subdisks.
VERY IMPORTANT: Remember not to associate any subdisk when the volume is started. By associating a subdisk when the volume is started, Volume Manager will recover the data to the subdisk and overwrite the subdisk. Always stop the volume before making any changes.
# vxvol -g stop
# vxsd -g dis
# vxsd -g -l / assoc
# vxmend -g fix unstale
For the above example, disk30-01 and disk55-01 are the next combination to be dissociated:
# vxvol -g stop r5vol
# vxsd -g dis disk55-01 # disk30-01 has already been dissociated
# vxsd -g -l 2/4096800 assoc r5vol-01 disk45-01
# vxmend -g fix unstale disk45-01
Then start the volume again and ask end-user to verify the data. In case of VxFS, get the VxFS image and fsck the VxFS image:
# vxvol -g -o delayrecover start
# metasave_5.8 /dev/vx/rdsk// | metasave_5.8 -r
# fsck -F vxfs -o full
Repeat the above procedure for all the combinations of subdisks until the end-user verifies the data to be correct.
Once the end-user verifies the data to be correct, the rest of the subdisks can be recovered by associating them back after the volume is started:
# vxsd -g -l / assoc
Raid5 logs failed while being zeroed / Plex contains unusable subdisk
--------------------------------------------------------------------------------------------------------
If the device for the log plex is not accessible, the RAID-5 volume can not be started even if it is forced to.
v r5vol - DISABLED CLEAN 24576000 raid5 - RAID
pl r5vol-01 r5vol DISABLED ACTIVE 24577920 RAID 5/32 RW
sd disk1-01 r5vol-01 disk1 0 2049120 0/0 c1t11d0 RCOV
sd disk10-01 r5vol-01 disk10 415 2047680 0/2049120 - NDEV
sd disk15-01 r5vol-01 disk15 415 2047680 0/4096800 sdc1t11d0s5 RCOV
sd disk3-01 r5vol-01 disk3 0 2049120 1/0 c1t13d0 ENA
sd disk30-01 r5vol-01 disk30 415 2047680 1/2049120 sdc1t13d0s0 ENA
sd disk35-01 r5vol-01 disk35 415 2047680 1/4096800 sdc1t13d0s5 ENA
sd disk4-01 r5vol-01 disk4 0 2049120 2/0 c1t14d0 ENA
sd disk40-01 r5vol-01 disk40 415 2047680 2/2049120 sdc1t14d0s0 ENA
sd disk45-01 r5vol-01 disk45 415 2047680 2/4096800 sdc1t14d0s5 ENA
sd disk5-01 r5vol-01 disk5 0 2049120 3/0 c1t15d0 ENA
sd disk50-01 r5vol-01 disk50 415 2047680 3/2049120 sdc1t15d0s0 ENA
sd disk55-01 r5vol-01 disk55 415 2047680 3/4096800 sdc1t15d0s5 ENA
sd disk6-01 r5vol-01 disk6 0 2049120 4/0 c1t6d0 ENA
sd disk60-01 r5vol-01 disk60 415 2047680 4/2049120 sdc1t6d0s0 ENA
sd disk65-01 r5vol-01 disk65 415 2047680 4/4096800 sdc1t6d0s5 ENA
pl r5vol-02 r5vol DISABLED BADLOG 5760 CONCAT - WO
sd disk56-01 r5vol-02 disk56 415 5760 0 sdc1t15d0s6 ENA <<<< Device sdc1t15d0s6 is not accessible
You can not even force this volume to start if the LOG device fails:
sydvcs4# vxvol -f -o delayrecover start r5vol
vxvm:vxvol: ERROR: Volume r5vol: Failed to zero logs offset 0 len 248
vxvm:vxvol: ERROR: No such file or directory
vxvm:vxvol: ERROR: Volume r5vol: Raid5 log(s) failed while being zeroed . The logs should be replaced before starting the volume
Another example is that the subdisk in the log plex is not mapped to physical device:
v r5vol - DISABLED ACTIVE 2457600 raid5 - RAID
pl r5vol-01 r5vol DISABLED ACTIVE 2459520 RAID 5/32 RW
sd disk6-01 r5vol-01 disk6 0 204960 0/0 c1t6d0 FAIL
sd disk60-01 r5vol-01 disk60 927 204960 0/204960 sdc1t6d0s0 FAIL
sd disk65-01 r5vol-01 disk65 927 204960 0/409920 sdc1t6d0s5 FAIL
sd disk1-01 r5vol-01 disk1 0 204960 1/0 c1t11d0 ENA
sd disk10-01 r5vol-01 disk10 927 204960 1/204960 sdc1t11d0s0 ENA
sd disk15-01 r5vol-01 disk15 927 204960 1/409920 sdc1t11d0s5 ENA
sd disk3-01 r5vol-01 disk3 0 204960 2/0 c1t13d0 ENA
sd disk30-01 r5vol-01 disk30 927 204960 2/204960 sdc1t13d0s0 ENA
sd disk35-01 r5vol-01 disk35 927 204960 2/409920 sdc1t13d0s5 ENA
sd disk4-01 r5vol-01 disk4 0 204960 3/0 c1t14d0 ENA
sd disk40-01 r5vol-01 disk40 927 204960 3/204960 sdc1t14d0s0 ENA
sd disk45-01 r5vol-01 disk45 927 204960 3/409920 sdc1t14d0s5 ENA
sd disk5-01 r5vol-01 disk5 0 204960 4/0 c1t15d0 ENA
sd disk50-01 r5vol-01 disk50 927 204960 4/204960 sdc1t15d0s0 RCOV
sd disk55-01 r5vol-01 disk55 927 204960 4/409920 sdc1t15d0s5 ENA
pl r5vol-02 r5vol DISABLED NODEVICE 5856 CONCAT - RW <<< Plex is marked as NODEVICE
sd disk56-01 r5vol-02 disk56 927 5856 0 - NDEV <<< Subdisk is not mapped into a device
# vxvol -f -o delayrecover start r5vol
vxvm:vxvol: ERROR: changing plex r5vol-02: Plex contains unusable subdisk
Try to recover the device before starting the volume. The log needs to be zeroed out before the volume can be started.
If you need to access the data urgently and you can not fix the log disk, you will need to dissociate the log plex and the volume will need to be forced to start
阅读(3627) | 评论(0) | 转发(0) |
