Have a master/client redhat linux server running Netbackup 8.2.
I Want to try and break up a current NDMP netbackup policy so that it kicks off groups of subdirectories one at a time using wildcards and with only one job at a time (limit jobs per policy = 1) .
With in our current NDMP policy we have the following "Backup Selections"
This volume is just over 50TB in size and Under the volume /client1/ we have like 12 subdirectories... All starting with alpha characters..
How can I use wildcards to break this volume up? or do I need to explicitly supply the full path.
I have tried this for example
/client/C* as there are 3 subdir that start with "C" but I get the following error
Error ndmpagent (pid=24809) readAllFSFromPath: No volumes found matching "/client/C*"
Also have tried :
/client/[A-C]* which also didnt work. Thinking I could grab any subdir starting with an A, B or C..
I guess if I have to list all the current subdir full paths out one by one its easy enough but thought wildcards could be used to some extent?.. If I did have to list each current subdir I guess I could create another policy to exclude the current subdirs so that it will catch new dirs that have been created. There isnt a lot of new subdirs being created but just in case it happens at least I would be covered.
The NDMP Admin guide for NetBackup provides more info...
Currently only NETAPP supports wildcards at folder level. For everything else (that is on the HCL) they only support wildcards at the VOLUME level.
The link below gives the HCL and lists the supported NDMP vendors. I don't see "NAS Filer" on it so it may be that you are trying to configure something that is NOT supported by Veritas (although it could still work).
Thanks StoneRam for the Links and the info.
So our NAS Device in which we are running this NDMP bkup is that of a DELL Compellent FS8600 environment.
So your saying wildcards only at the Volume level. So How could I possible use a wildcard at the volume level to spool out individual netbackup jobs for this voulme for each subdirectory... Is that even possible then.. Sounds like its not...
So Doing this at the volume level /Client1/* just going to give me one netbackup job running of all the subdirs and will bkup everything up.. Doing this /Client1/ is basically the same as above.
We just thought if there was a way to dump out each individual subdir under the volume as a seperate netbackup job at least we would know what has been backed up etc incase we need to stop the job for patching etc.. The way we have it now is that it just backs up using this /Client1/ as the backup selection of this policy and we have no idea what subdir it is on etc when the job is running... As I mention its roughly over 50TB in size so it runs for a long while..
I know I could list each subdir under the backup selections of this policy and maybe set the jobs per policy to like 2 to have 2 running at a time etc...
such as /client1/subdir1/, /client1/subdir2/....etc...../client1/subdir4/
This is feasible as there arent a ton of subdirs but the fear in doing it this way is that we have no control if someone goes out and creates a new subdir and if this is the case then it wouldnt be in the current netbackup policy until I added etc...
Yea there isn't a way for you to use the wildcards in the manner that you want with your NDMP filer. You might be able to setup one policy that explicitly list all of the current subdirectories minus 1.
Then setup another policy that would exclude all those subdirectories you listed in the other policy and also backup everything else (the 1 volume you didn't list in the other policy). Maybe something like this.
If any new subdirectories were created, the 2nd policy would catch them. I don't know if this would work but you could test it.
I think the "EXCLUDE" statements will work but If I use the ALL_FILESYSTEMS Its going to grab other Volumes (and their subdirs) we have on the NAS which I dont want. I quickly tested this and this is what happened.
Since the excludes work I got to find a way to backup the possible new subdirs in case if any are created..
While it is worth testing, I don't think the VOLUME_EXCLUDE_LIST will help as per the manual it applies at the volume level, not subdirectory level (I hope I'm wrong on this).
As for using the ALL_FILESYSTEMS, and grabbing other volumes, just add those additional volumes that are not /client1 to the exlcudes.
Have a look at this article which describes excluding directories from a NDMP backup for a EMC Celera Filer
Check out the documentation for your specific filer and see if there is something similar (it may well be the same directive). I can see that your filer has the ability to exclude directories, but not the specifics of how to achieve this (you could maybe ask your Dell SE for assistance on the directive to use).