Re: Recovering volumes that have been deleted?! - Veritas Volume Manager

This is a discussion on Re: Recovering volumes that have been deleted?! - Veritas Volume Manager ; Hi, Here's the vxprint out (/etc/vx/cbr does not exist, even though we have V3.5) v test200 - DISABLED CLEAN 419430400 SELECT - fsgen pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT - RW sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0 ...

+ Reply to Thread
Results 1 to 8 of 8

Thread: Re: Recovering volumes that have been deleted?!

  1. Re: Recovering volumes that have been deleted?!


    Hi,

    Here's the vxprint out (/etc/vx/cbr does not exist, even though we have V3.5)

    v test200 - DISABLED CLEAN 419430400 SELECT -
    fsgen
    pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    RW
    sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    ENA
    sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    ENA


    Kind Regards,
    Mark



  2. Re: Recovering volumes that have been deleted?!

    CBR came in with 3.5 MP1


    OK, so hetre we see that you have a simple CONCAT volume spanning 2
    sub-disks.

    So, the idea is to build the volume from the bottom up.
    (sub-disks, then plex, then volume)

    Remember that this will only work if you have not used the disks in
    question.

    Please look at the complete vxprint of this diskgroup, and look
    specifically for the disks "sandg01" and "sandg02"


    From this vxprint, we can see that the sub-disk (that is a part of the
    public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    unused is:
    from block 276824064 for another 40960 blocks



    Same goes for sandg02 (or disk c5t0d1)

    we need the area to be unused:
    from block 0 to block 419389440


    (a block is 512 bytes, so every 2 block will make 1KB)


    OK, once you've verified that this space is still available, we will use
    the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on
    sandg02) with the specified offsets and lengths. Also make sure that the
    subdisk names (sandg01-11 and sandg02-01) are not being used. If they
    are alraedy being used, you can just change these names in the commands
    below.



    I guess your diskgroup name here is "sandg"


    so here are the first commands:

    (in general the vxmake command for sub-disks are:
    vxmake -g sd sd-name,disk-anme,offset,length


    So, in your case for the 2 sub-diks
    vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    vxmake -g sandg sd sandg02-01,sandg02,0,419389440



    You can now check your handy-work with the vxprint command again, just
    to see if the subdisks we made.


    The next step is to create a plex. Now a plex consists of sub-disks.
    (the 2 new ones we just created). The biggest issue is that we need to
    known the layout of the plex (CONCAT in your case), and how the
    sub-disks fit into the plex. (not too difficult in your case. The secret
    again is to make sure that the plex name is not being used again (should
    not be used, unless you've got another volume called test200) and the
    sub-disk names we are going to use, are those we just created (so if you
    did not use the subdisk names I specified, just modify the command).

    The general form of the vxmake command for plexes are:

    vxmake -g plex plexname layout=
    sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......



    OK so in your case we will create plex test200-01, and we will use
    sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please
    note that the first subdisk is column 0 and the second sub-disk is
    column 40960 (well that is how far into the plex and volume the second
    sub-disk starts)


    vxmake -g sandg plex test200-01 layout=concat
    sd=sandg01-11:0,sd=sandg02-01:40960



    OK, that takes care of our plex, next the volume. It is important to
    specify a volume usage (general for all but raw and RAID5 volumes). We
    will also specify the length of the volume (419430400 in your case) and
    specify which plex to use (well, we've only created 1).

    The general vxmake command for volumes:

    vxmake -g -U vol len= plex=



    OK, so in your case:


    vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01



    OK, again, you can use the vxprint command to check your handy work. You
    should see something very similar to the vxprint that you posted. The
    bigegst issue, is that now we need to amke the volume start. There are a
    lot of steps that you can go through to make the plex clean, and then
    start the volume (etc.), but in yur case, because we have only a single
    plex, we can not mess it up, by just force starting it (don't ever do
    this for RAID5 volumes and for mirror volumes, please use the
    troubleshooting guide to enable 1 plex at a time - see if there is data
    on, before attaching the other plex).

    In your case, it would mean the following command:


    vxvol -g sandg -f start test200



    Again, check with vxprint, and now the volume should be ENABLE ACTIVE


    All you've got left to do, is check your data (fsck on the volume or mount)



    Now, sit back, drink beer and thank your stars that you've got the old
    vxprint output.



    hope that rings your bell !!! If you've got any other problems (like you
    don't know if the area on the disk are free, just post the whole vxprint
    here and I will have a look for you.

    Hope you've learned something as well. That is why I've gone into a biot
    more detail (rather than just give you the commands to run)



    Mark wrote:
    > Hi,
    >
    > Here's the vxprint out (/etc/vx/cbr does not exist, even though we have V3.5)
    >
    > v test200 - DISABLED CLEAN 419430400 SELECT -
    > fsgen
    > pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    > RW
    > sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    > ENA
    > sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    > ENA
    >
    >
    > Kind Regards,
    > Mark
    >
    >


  3. Re: Recovering volumes that have been deleted?!


    The subdisks are definitely not in use, and the space is still there, but
    when I try the vxmake, I get this error:


    "vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    vxvm:vxmake: ERROR: Bad subdisk name: name is too long"


    ...which is odd, because the other subdisks have this naming format...I can
    change the name, but it's a format that we use.


    Kind Regards,
    Mark


    Me wrote:
    >CBR came in with 3.5 MP1
    >
    >
    >OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >sub-disks.
    >
    >So, the idea is to build the volume from the bottom up.
    >(sub-disks, then plex, then volume)
    >
    >Remember that this will only work if you have not used the disks in
    >question.
    >
    >Please look at the complete vxprint of this diskgroup, and look
    >specifically for the disks "sandg01" and "sandg02"
    >
    >
    > From this vxprint, we can see that the sub-disk (that is a part of the


    >public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >unused is:
    >from block 276824064 for another 40960 blocks
    >
    >
    >
    >Same goes for sandg02 (or disk c5t0d1)
    >
    >we need the area to be unused:
    >from block 0 to block 419389440
    >
    >
    >(a block is 512 bytes, so every 2 block will make 1KB)
    >
    >
    >OK, once you've verified that this space is still available, we will use


    >the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on
    >sandg02) with the specified offsets and lengths. Also make sure that the


    >subdisk names (sandg01-11 and sandg02-01) are not being used. If they
    >are alraedy being used, you can just change these names in the commands


    >below.
    >
    >
    >
    >I guess your diskgroup name here is "sandg"
    >
    >
    >so here are the first commands:
    >
    >(in general the vxmake command for sub-disks are:
    > vxmake -g sd sd-name,disk-anme,offset,length
    >
    >
    >So, in your case for the 2 sub-diks
    >vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >
    >
    >
    >You can now check your handy-work with the vxprint command again, just
    >to see if the subdisks we made.
    >
    >
    >The next step is to create a plex. Now a plex consists of sub-disks.
    >(the 2 new ones we just created). The biggest issue is that we need to
    >known the layout of the plex (CONCAT in your case), and how the
    >sub-disks fit into the plex. (not too difficult in your case. The secret


    >again is to make sure that the plex name is not being used again (should


    >not be used, unless you've got another volume called test200) and the
    >sub-disk names we are going to use, are those we just created (so if you


    >did not use the subdisk names I specified, just modify the command).
    >
    >The general form of the vxmake command for plexes are:
    >
    >vxmake -g plex plexname layout=
    >sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >
    >
    >
    >OK so in your case we will create plex test200-01, and we will use
    >sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please
    >note that the first subdisk is column 0 and the second sub-disk is
    >column 40960 (well that is how far into the plex and volume the second
    >sub-disk starts)
    >
    >
    >vxmake -g sandg plex test200-01 layout=concat
    >sd=sandg01-11:0,sd=sandg02-01:40960
    >
    >
    >
    >OK, that takes care of our plex, next the volume. It is important to
    >specify a volume usage (general for all but raw and RAID5 volumes). We
    >will also specify the length of the volume (419430400 in your case) and


    >specify which plex to use (well, we've only created 1).
    >
    >The general vxmake command for volumes:
    >
    >vxmake -g -U vol len= plex=
    >
    >
    >
    >OK, so in your case:
    >
    >
    >vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >
    >
    >
    >OK, again, you can use the vxprint command to check your handy work. You


    >should see something very similar to the vxprint that you posted. The
    >bigegst issue, is that now we need to amke the volume start. There are a


    >lot of steps that you can go through to make the plex clean, and then
    >start the volume (etc.), but in yur case, because we have only a single


    >plex, we can not mess it up, by just force starting it (don't ever do
    >this for RAID5 volumes and for mirror volumes, please use the
    >troubleshooting guide to enable 1 plex at a time - see if there is data


    >on, before attaching the other plex).
    >
    >In your case, it would mean the following command:
    >
    >
    >vxvol -g sandg -f start test200
    >
    >
    >
    >Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >
    >
    >All you've got left to do, is check your data (fsck on the volume or mount)
    >
    >
    >
    >Now, sit back, drink beer and thank your stars that you've got the old
    >vxprint output.
    >
    >
    >
    >hope that rings your bell !!! If you've got any other problems (like you


    >don't know if the area on the disk are free, just post the whole vxprint


    >here and I will have a look for you.
    >
    >Hope you've learned something as well. That is why I've gone into a biot


    >more detail (rather than just give you the commands to run)
    >
    >
    >
    >Mark wrote:
    >> Hi,
    >>
    >> Here's the vxprint out (/etc/vx/cbr does not exist, even though we have

    V3.5)
    >>
    >> v test200 - DISABLED CLEAN 419430400 SELECT -
    >> fsgen
    >> pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >> RW
    >> sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >> ENA
    >> sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >> ENA
    >>
    >>
    >> Kind Regards,
    >> Mark
    >>
    >>



  4. Re: Recovering volumes that have been deleted?!


    The space is also still available, here's a vxdg free output:

    GROUP DISK DEVICE TAG OFFSET LENGTH FLAGS
    sandg sandg01 c5t0d0s2 c5t0d0 276824064 37707776 -
    sandg sandg02 c5t0d1s2 c5t0d1 0 419389440 -

    Cheers,
    Mark




    Me wrote:
    >CBR came in with 3.5 MP1
    >
    >
    >OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >sub-disks.
    >
    >So, the idea is to build the volume from the bottom up.
    >(sub-disks, then plex, then volume)
    >
    >Remember that this will only work if you have not used the disks in
    >question.
    >
    >Please look at the complete vxprint of this diskgroup, and look
    >specifically for the disks "sandg01" and "sandg02"
    >
    >
    > From this vxprint, we can see that the sub-disk (that is a part of the


    >public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >unused is:
    >from block 276824064 for another 40960 blocks
    >
    >
    >
    >Same goes for sandg02 (or disk c5t0d1)
    >
    >we need the area to be unused:
    >from block 0 to block 419389440
    >
    >
    >(a block is 512 bytes, so every 2 block will make 1KB)
    >
    >
    >OK, once you've verified that this space is still available, we will use


    >the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on
    >sandg02) with the specified offsets and lengths. Also make sure that the


    >subdisk names (sandg01-11 and sandg02-01) are not being used. If they
    >are alraedy being used, you can just change these names in the commands


    >below.
    >
    >
    >
    >I guess your diskgroup name here is "sandg"
    >
    >
    >so here are the first commands:
    >
    >(in general the vxmake command for sub-disks are:
    > vxmake -g sd sd-name,disk-anme,offset,length
    >
    >
    >So, in your case for the 2 sub-diks
    >vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >
    >
    >
    >You can now check your handy-work with the vxprint command again, just
    >to see if the subdisks we made.
    >
    >
    >The next step is to create a plex. Now a plex consists of sub-disks.
    >(the 2 new ones we just created). The biggest issue is that we need to
    >known the layout of the plex (CONCAT in your case), and how the
    >sub-disks fit into the plex. (not too difficult in your case. The secret


    >again is to make sure that the plex name is not being used again (should


    >not be used, unless you've got another volume called test200) and the
    >sub-disk names we are going to use, are those we just created (so if you


    >did not use the subdisk names I specified, just modify the command).
    >
    >The general form of the vxmake command for plexes are:
    >
    >vxmake -g plex plexname layout=
    >sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >
    >
    >
    >OK so in your case we will create plex test200-01, and we will use
    >sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please
    >note that the first subdisk is column 0 and the second sub-disk is
    >column 40960 (well that is how far into the plex and volume the second
    >sub-disk starts)
    >
    >
    >vxmake -g sandg plex test200-01 layout=concat
    >sd=sandg01-11:0,sd=sandg02-01:40960
    >
    >
    >
    >OK, that takes care of our plex, next the volume. It is important to
    >specify a volume usage (general for all but raw and RAID5 volumes). We
    >will also specify the length of the volume (419430400 in your case) and


    >specify which plex to use (well, we've only created 1).
    >
    >The general vxmake command for volumes:
    >
    >vxmake -g -U vol len= plex=
    >
    >
    >
    >OK, so in your case:
    >
    >
    >vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >
    >
    >
    >OK, again, you can use the vxprint command to check your handy work. You


    >should see something very similar to the vxprint that you posted. The
    >bigegst issue, is that now we need to amke the volume start. There are a


    >lot of steps that you can go through to make the plex clean, and then
    >start the volume (etc.), but in yur case, because we have only a single


    >plex, we can not mess it up, by just force starting it (don't ever do
    >this for RAID5 volumes and for mirror volumes, please use the
    >troubleshooting guide to enable 1 plex at a time - see if there is data


    >on, before attaching the other plex).
    >
    >In your case, it would mean the following command:
    >
    >
    >vxvol -g sandg -f start test200
    >
    >
    >
    >Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >
    >
    >All you've got left to do, is check your data (fsck on the volume or mount)
    >
    >
    >
    >Now, sit back, drink beer and thank your stars that you've got the old
    >vxprint output.
    >
    >
    >
    >hope that rings your bell !!! If you've got any other problems (like you


    >don't know if the area on the disk are free, just post the whole vxprint


    >here and I will have a look for you.
    >
    >Hope you've learned something as well. That is why I've gone into a biot


    >more detail (rather than just give you the commands to run)
    >
    >
    >
    >Mark wrote:
    >> Hi,
    >>
    >> Here's the vxprint out (/etc/vx/cbr does not exist, even though we have

    V3.5)
    >>
    >> v test200 - DISABLED CLEAN 419430400 SELECT -
    >> fsgen
    >> pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >> RW
    >> sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >> ENA
    >> sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >> ENA
    >>
    >>
    >> Kind Regards,
    >> Mark
    >>
    >>



  5. Re: Recovering volumes that have been deleted?!


    Hi,

    I worked out the problem of the subdisk name, there's no "," between the
    sub-disk name and diskname ! Managed to go through the rest of it, and voila
    - my filesystem is back with all it's data. Thank you very much for all your
    help, and yes I have learnt a lot as well, I'll keep your email for future
    reference (but hopefully I won't need to do that again !)

    Thanks again.

    Mark


    Me wrote:
    >CBR came in with 3.5 MP1
    >
    >
    >OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >sub-disks.
    >
    >So, the idea is to build the volume from the bottom up.
    >(sub-disks, then plex, then volume)
    >
    >Remember that this will only work if you have not used the disks in
    >question.
    >
    >Please look at the complete vxprint of this diskgroup, and look
    >specifically for the disks "sandg01" and "sandg02"
    >
    >
    > From this vxprint, we can see that the sub-disk (that is a part of the


    >public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >unused is:
    >from block 276824064 for another 40960 blocks
    >
    >
    >
    >Same goes for sandg02 (or disk c5t0d1)
    >
    >we need the area to be unused:
    >from block 0 to block 419389440
    >
    >
    >(a block is 512 bytes, so every 2 block will make 1KB)
    >
    >
    >OK, once you've verified that this space is still available, we will use


    >the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on
    >sandg02) with the specified offsets and lengths. Also make sure that the


    >subdisk names (sandg01-11 and sandg02-01) are not being used. If they
    >are alraedy being used, you can just change these names in the commands


    >below.
    >
    >
    >
    >I guess your diskgroup name here is "sandg"
    >
    >
    >so here are the first commands:
    >
    >(in general the vxmake command for sub-disks are:
    > vxmake -g sd sd-name,disk-anme,offset,length
    >
    >
    >So, in your case for the 2 sub-diks
    >vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >
    >
    >
    >You can now check your handy-work with the vxprint command again, just
    >to see if the subdisks we made.
    >
    >
    >The next step is to create a plex. Now a plex consists of sub-disks.
    >(the 2 new ones we just created). The biggest issue is that we need to
    >known the layout of the plex (CONCAT in your case), and how the
    >sub-disks fit into the plex. (not too difficult in your case. The secret


    >again is to make sure that the plex name is not being used again (should


    >not be used, unless you've got another volume called test200) and the
    >sub-disk names we are going to use, are those we just created (so if you


    >did not use the subdisk names I specified, just modify the command).
    >
    >The general form of the vxmake command for plexes are:
    >
    >vxmake -g plex plexname layout=
    >sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >
    >
    >
    >OK so in your case we will create plex test200-01, and we will use
    >sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please
    >note that the first subdisk is column 0 and the second sub-disk is
    >column 40960 (well that is how far into the plex and volume the second
    >sub-disk starts)
    >
    >
    >vxmake -g sandg plex test200-01 layout=concat
    >sd=sandg01-11:0,sd=sandg02-01:40960
    >
    >
    >
    >OK, that takes care of our plex, next the volume. It is important to
    >specify a volume usage (general for all but raw and RAID5 volumes). We
    >will also specify the length of the volume (419430400 in your case) and


    >specify which plex to use (well, we've only created 1).
    >
    >The general vxmake command for volumes:
    >
    >vxmake -g -U vol len= plex=
    >
    >
    >
    >OK, so in your case:
    >
    >
    >vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >
    >
    >
    >OK, again, you can use the vxprint command to check your handy work. You


    >should see something very similar to the vxprint that you posted. The
    >bigegst issue, is that now we need to amke the volume start. There are a


    >lot of steps that you can go through to make the plex clean, and then
    >start the volume (etc.), but in yur case, because we have only a single


    >plex, we can not mess it up, by just force starting it (don't ever do
    >this for RAID5 volumes and for mirror volumes, please use the
    >troubleshooting guide to enable 1 plex at a time - see if there is data


    >on, before attaching the other plex).
    >
    >In your case, it would mean the following command:
    >
    >
    >vxvol -g sandg -f start test200
    >
    >
    >
    >Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >
    >
    >All you've got left to do, is check your data (fsck on the volume or mount)
    >
    >
    >
    >Now, sit back, drink beer and thank your stars that you've got the old
    >vxprint output.
    >
    >
    >
    >hope that rings your bell !!! If you've got any other problems (like you


    >don't know if the area on the disk are free, just post the whole vxprint


    >here and I will have a look for you.
    >
    >Hope you've learned something as well. That is why I've gone into a biot


    >more detail (rather than just give you the commands to run)
    >
    >
    >
    >Mark wrote:
    >> Hi,
    >>
    >> Here's the vxprint out (/etc/vx/cbr does not exist, even though we have

    V3.5)
    >>
    >> v test200 - DISABLED CLEAN 419430400 SELECT -
    >> fsgen
    >> pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >> RW
    >> sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >> ENA
    >> sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >> ENA
    >>
    >>
    >> Kind Regards,
    >> Mark
    >>
    >>



  6. Re: Recovering volumes that have been deleted?!

    Sorry, typo on the ","

    You were lucky in that you did not do anything asfter deleting the
    volume, and the fact that you had an old "vxprint" output. (Good idea
    too keep one)


    Mark wrote:
    > Hi,
    >
    > I worked out the problem of the subdisk name, there's no "," between the
    > sub-disk name and diskname ! Managed to go through the rest of it, and voila
    > - my filesystem is back with all it's data. Thank you very much for all your
    > help, and yes I have learnt a lot as well, I'll keep your email for future
    > reference (but hopefully I won't need to do that again !)
    >
    > Thanks again.
    >
    > Mark
    >
    >
    > Me wrote:
    >
    >>CBR came in with 3.5 MP1
    >>
    >>
    >>OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >>sub-disks.
    >>
    >>So, the idea is to build the volume from the bottom up.
    >>(sub-disks, then plex, then volume)
    >>
    >>Remember that this will only work if you have not used the disks in
    >>question.
    >>
    >>Please look at the complete vxprint of this diskgroup, and look
    >>specifically for the disks "sandg01" and "sandg02"
    >>
    >>
    >>From this vxprint, we can see that the sub-disk (that is a part of the

    >
    >
    >>public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >>unused is:

    >
    >>from block 276824064 for another 40960 blocks

    >
    >>
    >>
    >>Same goes for sandg02 (or disk c5t0d1)
    >>
    >>we need the area to be unused:

    >
    >>from block 0 to block 419389440

    >
    >>
    >>(a block is 512 bytes, so every 2 block will make 1KB)
    >>
    >>
    >>OK, once you've verified that this space is still available, we will use

    >
    >
    >>the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on
    >>sandg02) with the specified offsets and lengths. Also make sure that the

    >
    >
    >>subdisk names (sandg01-11 and sandg02-01) are not being used. If they
    >>are alraedy being used, you can just change these names in the commands

    >
    >
    >>below.
    >>
    >>
    >>
    >>I guess your diskgroup name here is "sandg"
    >>
    >>
    >>so here are the first commands:
    >>
    >>(in general the vxmake command for sub-disks are:
    >> vxmake -g sd sd-name,disk-anme,offset,length
    >>
    >>
    >>So, in your case for the 2 sub-diks
    >>vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >>vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >>
    >>
    >>
    >>You can now check your handy-work with the vxprint command again, just
    >>to see if the subdisks we made.
    >>
    >>
    >>The next step is to create a plex. Now a plex consists of sub-disks.
    >>(the 2 new ones we just created). The biggest issue is that we need to
    >>known the layout of the plex (CONCAT in your case), and how the
    >>sub-disks fit into the plex. (not too difficult in your case. The secret

    >
    >
    >>again is to make sure that the plex name is not being used again (should

    >
    >
    >>not be used, unless you've got another volume called test200) and the
    >>sub-disk names we are going to use, are those we just created (so if you

    >
    >
    >>did not use the subdisk names I specified, just modify the command).
    >>
    >>The general form of the vxmake command for plexes are:
    >>
    >>vxmake -g plex plexname layout=
    >>sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >>
    >>
    >>
    >>OK so in your case we will create plex test200-01, and we will use
    >>sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please
    >>note that the first subdisk is column 0 and the second sub-disk is
    >>column 40960 (well that is how far into the plex and volume the second
    >>sub-disk starts)
    >>
    >>
    >>vxmake -g sandg plex test200-01 layout=concat
    >>sd=sandg01-11:0,sd=sandg02-01:40960
    >>
    >>
    >>
    >>OK, that takes care of our plex, next the volume. It is important to
    >>specify a volume usage (general for all but raw and RAID5 volumes). We
    >>will also specify the length of the volume (419430400 in your case) and

    >
    >
    >>specify which plex to use (well, we've only created 1).
    >>
    >>The general vxmake command for volumes:
    >>
    >>vxmake -g -U vol len= plex=
    >>
    >>
    >>
    >>OK, so in your case:
    >>
    >>
    >>vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >>
    >>
    >>
    >>OK, again, you can use the vxprint command to check your handy work. You

    >
    >
    >>should see something very similar to the vxprint that you posted. The
    >>bigegst issue, is that now we need to amke the volume start. There are a

    >
    >
    >>lot of steps that you can go through to make the plex clean, and then
    >>start the volume (etc.), but in yur case, because we have only a single

    >
    >
    >>plex, we can not mess it up, by just force starting it (don't ever do
    >>this for RAID5 volumes and for mirror volumes, please use the
    >>troubleshooting guide to enable 1 plex at a time - see if there is data

    >
    >
    >>on, before attaching the other plex).
    >>
    >>In your case, it would mean the following command:
    >>
    >>
    >>vxvol -g sandg -f start test200
    >>
    >>
    >>
    >>Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >>
    >>
    >>All you've got left to do, is check your data (fsck on the volume or mount)
    >>
    >>
    >>
    >>Now, sit back, drink beer and thank your stars that you've got the old
    >>vxprint output.
    >>
    >>
    >>
    >>hope that rings your bell !!! If you've got any other problems (like you

    >
    >
    >>don't know if the area on the disk are free, just post the whole vxprint

    >
    >
    >>here and I will have a look for you.
    >>
    >>Hope you've learned something as well. That is why I've gone into a biot

    >
    >
    >>more detail (rather than just give you the commands to run)
    >>
    >>
    >>
    >>Mark wrote:
    >>
    >>>Hi,
    >>>
    >>>Here's the vxprint out (/etc/vx/cbr does not exist, even though we have

    >
    > V3.5)
    >
    >>>v test200 - DISABLED CLEAN 419430400 SELECT -
    >>>fsgen
    >>>pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >>>RW
    >>>sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >>>ENA
    >>>sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >>>ENA
    >>>
    >>>
    >>>Kind Regards,
    >>>Mark
    >>>
    >>>

    >
    >


  7. Re: Recovering volumes that have been deleted?!


    When I took a earlier Volume Manager class ...

    A slide said that you can take the output of the vxprint and use this as
    input to vxmake to restore the volume. What is the proper command sequence
    along with all the options needed to recover a volume? Is this available
    for 3.5 and later versions for Solaris?

    My class book also mentions that the saved vxprint output may have to edited
    to remove DG records, can this be avoided?

    Paul


    Me wrote:
    >Sorry, typo on the ","
    >
    >You were lucky in that you did not do anything asfter deleting the
    >volume, and the fact that you had an old "vxprint" output. (Good idea
    >too keep one)
    >
    >
    >Mark wrote:
    >> Hi,
    >>
    >> I worked out the problem of the subdisk name, there's no "," between the
    >> sub-disk name and diskname ! Managed to go through the rest of it, and

    voila
    >> - my filesystem is back with all it's data. Thank you very much for all

    your
    >> help, and yes I have learnt a lot as well, I'll keep your email for future
    >> reference (but hopefully I won't need to do that again !)
    >>
    >> Thanks again.
    >>
    >> Mark
    >>
    >>
    >> Me wrote:
    >>
    >>>CBR came in with 3.5 MP1
    >>>
    >>>
    >>>OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >>>sub-disks.
    >>>
    >>>So, the idea is to build the volume from the bottom up.
    >>>(sub-disks, then plex, then volume)
    >>>
    >>>Remember that this will only work if you have not used the disks in
    >>>question.
    >>>
    >>>Please look at the complete vxprint of this diskgroup, and look
    >>>specifically for the disks "sandg01" and "sandg02"
    >>>
    >>>
    >>>From this vxprint, we can see that the sub-disk (that is a part of the

    >>
    >>
    >>>public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >>>unused is:

    >>
    >>>from block 276824064 for another 40960 blocks

    >>
    >>>
    >>>
    >>>Same goes for sandg02 (or disk c5t0d1)
    >>>
    >>>we need the area to be unused:

    >>
    >>>from block 0 to block 419389440

    >>
    >>>
    >>>(a block is 512 bytes, so every 2 block will make 1KB)
    >>>
    >>>
    >>>OK, once you've verified that this space is still available, we will use

    >>
    >>
    >>>the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on


    >>>sandg02) with the specified offsets and lengths. Also make sure that the

    >>
    >>
    >>>subdisk names (sandg01-11 and sandg02-01) are not being used. If they


    >>>are alraedy being used, you can just change these names in the commands

    >>
    >>
    >>>below.
    >>>
    >>>
    >>>
    >>>I guess your diskgroup name here is "sandg"
    >>>
    >>>
    >>>so here are the first commands:
    >>>
    >>>(in general the vxmake command for sub-disks are:
    >>> vxmake -g sd sd-name,disk-anme,offset,length
    >>>
    >>>
    >>>So, in your case for the 2 sub-diks
    >>>vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >>>vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >>>
    >>>
    >>>
    >>>You can now check your handy-work with the vxprint command again, just


    >>>to see if the subdisks we made.
    >>>
    >>>
    >>>The next step is to create a plex. Now a plex consists of sub-disks.
    >>>(the 2 new ones we just created). The biggest issue is that we need to


    >>>known the layout of the plex (CONCAT in your case), and how the
    >>>sub-disks fit into the plex. (not too difficult in your case. The secret

    >>
    >>
    >>>again is to make sure that the plex name is not being used again (should

    >>
    >>
    >>>not be used, unless you've got another volume called test200) and the


    >>>sub-disk names we are going to use, are those we just created (so if you

    >>
    >>
    >>>did not use the subdisk names I specified, just modify the command).
    >>>
    >>>The general form of the vxmake command for plexes are:
    >>>
    >>>vxmake -g plex plexname layout=
    >>>sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >>>
    >>>
    >>>
    >>>OK so in your case we will create plex test200-01, and we will use
    >>>sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please


    >>>note that the first subdisk is column 0 and the second sub-disk is
    >>>column 40960 (well that is how far into the plex and volume the second


    >>>sub-disk starts)
    >>>
    >>>
    >>>vxmake -g sandg plex test200-01 layout=concat
    >>>sd=sandg01-11:0,sd=sandg02-01:40960
    >>>
    >>>
    >>>
    >>>OK, that takes care of our plex, next the volume. It is important to
    >>>specify a volume usage (general for all but raw and RAID5 volumes). We


    >>>will also specify the length of the volume (419430400 in your case) and

    >>
    >>
    >>>specify which plex to use (well, we've only created 1).
    >>>
    >>>The general vxmake command for volumes:
    >>>
    >>>vxmake -g -U vol len= plex=
    >>>
    >>>
    >>>
    >>>OK, so in your case:
    >>>
    >>>
    >>>vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >>>
    >>>
    >>>
    >>>OK, again, you can use the vxprint command to check your handy work. You

    >>
    >>
    >>>should see something very similar to the vxprint that you posted. The


    >>>bigegst issue, is that now we need to amke the volume start. There are

    a
    >>
    >>
    >>>lot of steps that you can go through to make the plex clean, and then


    >>>start the volume (etc.), but in yur case, because we have only a single

    >>
    >>
    >>>plex, we can not mess it up, by just force starting it (don't ever do


    >>>this for RAID5 volumes and for mirror volumes, please use the
    >>>troubleshooting guide to enable 1 plex at a time - see if there is data

    >>
    >>
    >>>on, before attaching the other plex).
    >>>
    >>>In your case, it would mean the following command:
    >>>
    >>>
    >>>vxvol -g sandg -f start test200
    >>>
    >>>
    >>>
    >>>Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >>>
    >>>
    >>>All you've got left to do, is check your data (fsck on the volume or mount)
    >>>
    >>>
    >>>
    >>>Now, sit back, drink beer and thank your stars that you've got the old


    >>>vxprint output.
    >>>
    >>>
    >>>
    >>>hope that rings your bell !!! If you've got any other problems (like you

    >>
    >>
    >>>don't know if the area on the disk are free, just post the whole vxprint

    >>
    >>
    >>>here and I will have a look for you.
    >>>
    >>>Hope you've learned something as well. That is why I've gone into a biot

    >>
    >>
    >>>more detail (rather than just give you the commands to run)
    >>>
    >>>
    >>>
    >>>Mark wrote:
    >>>
    >>>>Hi,
    >>>>
    >>>>Here's the vxprint out (/etc/vx/cbr does not exist, even though we have

    >>
    >> V3.5)
    >>
    >>>>v test200 - DISABLED CLEAN 419430400 SELECT -
    >>>>fsgen
    >>>>pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >>>>RW
    >>>>sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >>>>ENA
    >>>>sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >>>>ENA
    >>>>
    >>>>
    >>>>Kind Regards,
    >>>>Mark
    >>>>
    >>>>

    >>
    >>



  8. Re: Recovering volumes that have been deleted?!

    vxprint -g -mdlpqQv > makefile

    make -g -d

    Paul Rontanini wrote:
    > When I took a earlier Volume Manager class ...
    >
    > A slide said that you can take the output of the vxprint and use this as
    > input to vxmake to restore the volume. What is the proper command sequence
    > along with all the options needed to recover a volume? Is this available
    > for 3.5 and later versions for Solaris?
    >
    > My class book also mentions that the saved vxprint output may have to edited
    > to remove DG records, can this be avoided?
    >
    > Paul
    >
    >
    > Me wrote:
    >
    >>Sorry, typo on the ","
    >>
    >>You were lucky in that you did not do anything asfter deleting the
    >>volume, and the fact that you had an old "vxprint" output. (Good idea
    >>too keep one)
    >>
    >>
    >>Mark wrote:
    >>
    >>>Hi,
    >>>
    >>>I worked out the problem of the subdisk name, there's no "," between the
    >>>sub-disk name and diskname ! Managed to go through the rest of it, and

    >
    > voila
    >
    >>>- my filesystem is back with all it's data. Thank you very much for all

    >
    > your
    >
    >>>help, and yes I have learnt a lot as well, I'll keep your email for future
    >>>reference (but hopefully I won't need to do that again !)
    >>>
    >>>Thanks again.
    >>>
    >>>Mark
    >>>
    >>>
    >>>Me wrote:
    >>>
    >>>
    >>>>CBR came in with 3.5 MP1
    >>>>
    >>>>
    >>>>OK, so hetre we see that you have a simple CONCAT volume spanning 2
    >>>>sub-disks.
    >>>>
    >>>>So, the idea is to build the volume from the bottom up.
    >>>>(sub-disks, then plex, then volume)
    >>>>
    >>>>Remember that this will only work if you have not used the disks in
    >>>>question.
    >>>>
    >>>>Please look at the complete vxprint of this diskgroup, and look
    >>>>specifically for the disks "sandg01" and "sandg02"
    >>>>
    >>>>
    >>>
    >>>>From this vxprint, we can see that the sub-disk (that is a part of the
    >>>
    >>>
    >>>
    >>>>public slice on the disk) for sandg01 (or c5t0d0) that needs to be
    >>>>unused is:
    >>>
    >>>>from block 276824064 for another 40960 blocks
    >>>
    >>>
    >>>>
    >>>>Same goes for sandg02 (or disk c5t0d1)
    >>>>
    >>>>we need the area to be unused:
    >>>
    >>>>from block 0 to block 419389440
    >>>
    >>>
    >>>>(a block is 512 bytes, so every 2 block will make 1KB)
    >>>>
    >>>>
    >>>>OK, once you've verified that this space is still available, we will use
    >>>
    >>>
    >>>>the vxmake comamnd to make 2 sub-disks (on on sandg01 and the other on

    >
    >
    >>>>sandg02) with the specified offsets and lengths. Also make sure that the
    >>>
    >>>
    >>>>subdisk names (sandg01-11 and sandg02-01) are not being used. If they

    >
    >
    >>>>are alraedy being used, you can just change these names in the commands
    >>>
    >>>
    >>>>below.
    >>>>
    >>>>
    >>>>
    >>>>I guess your diskgroup name here is "sandg"
    >>>>
    >>>>
    >>>>so here are the first commands:
    >>>>
    >>>>(in general the vxmake command for sub-disks are:
    >>>> vxmake -g sd sd-name,disk-anme,offset,length
    >>>>
    >>>>
    >>>>So, in your case for the 2 sub-diks
    >>>>vxmake -g sandg sd sandg01-11,sandg01,276824064,40960
    >>>>vxmake -g sandg sd sandg02-01,sandg02,0,419389440
    >>>>
    >>>>
    >>>>
    >>>>You can now check your handy-work with the vxprint command again, just

    >
    >
    >>>>to see if the subdisks we made.
    >>>>
    >>>>
    >>>>The next step is to create a plex. Now a plex consists of sub-disks.
    >>>>(the 2 new ones we just created). The biggest issue is that we need to

    >
    >
    >>>>known the layout of the plex (CONCAT in your case), and how the
    >>>>sub-disks fit into the plex. (not too difficult in your case. The secret
    >>>
    >>>
    >>>>again is to make sure that the plex name is not being used again (should
    >>>
    >>>
    >>>>not be used, unless you've got another volume called test200) and the

    >
    >
    >>>>sub-disk names we are going to use, are those we just created (so if you
    >>>
    >>>
    >>>>did not use the subdisk names I specified, just modify the command).
    >>>>
    >>>>The general form of the vxmake command for plexes are:
    >>>>
    >>>>vxmake -g plex plexname layout=
    >>>>sd=sub-disk_name:column,sd=sub-disk_name:column,sd=.......
    >>>>
    >>>>
    >>>>
    >>>>OK so in your case we will create plex test200-01, and we will use
    >>>>sandg01-11 and sandg02-01 as ou sub-disks with a CONCAT layout. Please

    >
    >
    >>>>note that the first subdisk is column 0 and the second sub-disk is
    >>>>column 40960 (well that is how far into the plex and volume the second

    >
    >
    >>>>sub-disk starts)
    >>>>
    >>>>
    >>>>vxmake -g sandg plex test200-01 layout=concat
    >>>>sd=sandg01-11:0,sd=sandg02-01:40960
    >>>>
    >>>>
    >>>>
    >>>>OK, that takes care of our plex, next the volume. It is important to
    >>>>specify a volume usage (general for all but raw and RAID5 volumes). We

    >
    >
    >>>>will also specify the length of the volume (419430400 in your case) and
    >>>
    >>>
    >>>>specify which plex to use (well, we've only created 1).
    >>>>
    >>>>The general vxmake command for volumes:
    >>>>
    >>>>vxmake -g -U vol len= plex=
    >>>>
    >>>>
    >>>>
    >>>>OK, so in your case:
    >>>>
    >>>>
    >>>>vxmake -g sandg -U gen vol test200 len=419430400 plex=test200-01
    >>>>
    >>>>
    >>>>
    >>>>OK, again, you can use the vxprint command to check your handy work. You
    >>>
    >>>
    >>>>should see something very similar to the vxprint that you posted. The

    >
    >
    >>>>bigegst issue, is that now we need to amke the volume start. There are

    >
    > a
    >
    >>>
    >>>>lot of steps that you can go through to make the plex clean, and then

    >
    >
    >>>>start the volume (etc.), but in yur case, because we have only a single
    >>>
    >>>
    >>>>plex, we can not mess it up, by just force starting it (don't ever do

    >
    >
    >>>>this for RAID5 volumes and for mirror volumes, please use the
    >>>>troubleshooting guide to enable 1 plex at a time - see if there is data
    >>>
    >>>
    >>>>on, before attaching the other plex).
    >>>>
    >>>>In your case, it would mean the following command:
    >>>>
    >>>>
    >>>>vxvol -g sandg -f start test200
    >>>>
    >>>>
    >>>>
    >>>>Again, check with vxprint, and now the volume should be ENABLE ACTIVE
    >>>>
    >>>>
    >>>>All you've got left to do, is check your data (fsck on the volume or mount)
    >>>>
    >>>>
    >>>>
    >>>>Now, sit back, drink beer and thank your stars that you've got the old

    >
    >
    >>>>vxprint output.
    >>>>
    >>>>
    >>>>
    >>>>hope that rings your bell !!! If you've got any other problems (like you
    >>>
    >>>
    >>>>don't know if the area on the disk are free, just post the whole vxprint
    >>>
    >>>
    >>>>here and I will have a look for you.
    >>>>
    >>>>Hope you've learned something as well. That is why I've gone into a biot
    >>>
    >>>
    >>>>more detail (rather than just give you the commands to run)
    >>>>
    >>>>
    >>>>
    >>>>Mark wrote:
    >>>>
    >>>>
    >>>>>Hi,
    >>>>>
    >>>>>Here's the vxprint out (/etc/vx/cbr does not exist, even though we have
    >>>
    >>>V3.5)
    >>>
    >>>
    >>>>>v test200 - DISABLED CLEAN 419430400 SELECT -
    >>>>>fsgen
    >>>>>pl test200-01 test200 DISABLED CLEAN 419430400 CONCAT -
    >>>>>RW
    >>>>>sd sandg01-11 test200-01 sandg01 276824064 40960 0 c5t0d0
    >>>>>ENA
    >>>>>sd sandg02-01 test200-01 sandg02 0 419389440 40960 c5t0d1
    >>>>>ENA
    >>>>>
    >>>>>
    >>>>>Kind Regards,
    >>>>>Mark
    >>>>>
    >>>>>
    >>>
    >>>

    >


+ Reply to Thread