tape in use question

[virag0640]

Hi,
I have configured some very long running NDMP jobs
using DDS. The initial job is working fine,
now I am going to setup 2 more similar ones on
2 more NAS servers.

However I am unsure about how tapes will be used.

I have created an NDMP pool for the backups with
a number of volumes in it.

When I monitor the current production job, I get
different results deppending on how I query the volumes:

# mminfo -av 105191
volume client date time size ssid fl lvl name
105191 my_ndmp_srv 04/05/08 06:00:16 PM 234 GB 1864026770 mi full /root_vdm_1/fsrv1_staff

As you can see, the flags show "mi" which means the
backup is "in progress" and spans volumes.

But when I query the volume another way:
mminfo -av -r "volume,state,pool,location,volretent" 105191
volume pool location expires
105191 NDMP Weekly L700 undef

It shows expiration as "undef". Now my question is,
is Legato smart enough to know not to use volumes that have
spanned data on them, when the job is "in progress"?

IE, if the job is not complete, another task coming along
behind for one of my other NDMP backups is not likely
to grab the tape and utilise it, thereby making my
first job have incomplete data?

I know this sounds like a stupid question, but I
actually don't know how Legato allocates volumes
on what basis?


cheers
rachel

 

[605]

Hi Rachel,

please see below ....

Now my question is, is Legato smart enough to know not to use volumes that have spanned data on them, when the job is "in progress"?

>> Not really. If a new job is started AND the volume fits AND the target session/max. parallelism would allow that, NW will also use this tape right away.


IE, if the job is not complete, another task coming along
behind for one of my other NDMP backups is not likely
to grab the tape and utilise it, thereby making my
first job have incomplete data?

>> With NDMP the situation is different as NDMP does not allow multplexing. So in this case, a new job must use another media on another device (if possible). Otherwise, it will be queued.

Als long as there is no error, NW will never abort a job by itself.


I know this sounds like a stupid question, but I
actually don't know how Legato allocates volumes
on what basis?

>> There is a coarse description in the manuals. However, it can just be rough as a lot of other conditions are also possible.