I know normally it is considered safer to have one pool that saves indexes & bootstrap for all backups
Actually it is believed to be that way, but it is not safer at all. It is easier for management I would rather say.
We tried and used the Storage node hostname under the client resoruce for the backup server along with nsrserverhost. This helped write index in the same tape where data was being written on the storage node but it still keeps waiting for a media for writing the bootstrap.
Bootstrap will be saved always on server. Index backup on storage node seems to be silently enabled in some 7.x version. I myself have no idea if that is ok or bug (recently I have seen variant of it which is certainly bug), but I dislike it. Always, and I mean always, index and bootstrap were data which originally reside on server - thus owned by server. Logically, they should be saved to storage (disk or tape) assigned to server. To have storage node save index (which is located on server) is not something I prefer or even wish to consider unless index db will be based on storage node in that case (which is not). A bug I have seen is where index gets saved under ownership of client - however that is really under specific circumstances and environment I doubt anyone else has.
Is there a logical way out i.e. I want to write idnex & bootstrap also on the same media on which I wirte my backups when writing a backup on the storage node?
Same media I doubt, but same pool yes. I doubt about same media as you get immediate request for tape for pool X on server while your tape device with tape from pool X on storage node needs to shift from writing,idle to writing,done state. NW won't wait for that of course (which is fine) and will load another tape from that pool (or Default if pool X can't be used on server) and use it.
Actually it is believed to be that way, but it is not safer at all. It is easier for management I would rather say.
I agree but it goes as a practive whenevr we want to use a Storage node - direct or dedicated.
Bootstrap will be saved always on server. Index backup on storage node seems to be silently enabled in some 7.x version. I myself have no idea if that is ok or bug (recently I have seen variant of it which is certainly bug), but I dislike it. Always, and I mean always, index and bootstrap were data which originally reside on server - thus owned by server. Logically, they should be saved to storage (disk or tape) assigned to server. To have storage node save index (which is located on server) is not something I prefer or even wish to consider unless index db will be based on storage node in that case (which is not). A bug I have seen is where index gets saved under ownership of client - however that is really under specific circumstances and environment I doubt anyone else has.
Yes but I tried this on 6.x which I know is unsupported but worked well here.
Same media I doubt, but same pool yes. I doubt about same media as you get immediate request for tape for pool X on server while your tape device with tape from pool X on storage node needs to shift from writing,idle to writing,done state. NW won't wait for that of course (which is fine) and will load another tape from that pool (or Default if pool X can't be used on server) and use it.
Does this suggest I should then have a media labelled for that pool on both the server as well as storage node?
I checked in 7.3.2 admin guide that says "A bootstrap cannot be backed up to the remote device".
ble1
4 Operator
•
14.4K Posts
1
February 21st, 2007 10:00
pool that saves indexes & bootstrap for all backups
Actually it is believed to be that way, but it is not safer at all. It is easier for management I would rather say.
client resoruce for the backup server along with
nsrserverhost. This helped write index in the same
tape where data was being written on the storage node
but it still keeps waiting for a media for writing
the bootstrap.
Bootstrap will be saved always on server. Index backup on storage node seems to be silently enabled in some 7.x version. I myself have no idea if that is ok or bug (recently I have seen variant of it which is certainly bug), but I dislike it. Always, and I mean always, index and bootstrap were data which originally reside on server - thus owned by server. Logically, they should be saved to storage (disk or tape) assigned to server. To have storage node save index (which is located on server) is not something I prefer or even wish to consider unless index db will be based on storage node in that case (which is not). A bug I have seen is where index gets saved under ownership of client - however that is really under specific circumstances and environment I doubt anyone else has.
& bootstrap also on the same media on which I wirte
my backups when writing a backup on the storage node?
Same media I doubt, but same pool yes. I doubt about same media as you get immediate request for tape for pool X on server while your tape device with tape from pool X on storage node needs to shift from writing,idle to writing,done state. NW won't wait for that of course (which is fine) and will load another tape from that pool (or Default if pool X can't be used on server) and use it.
amediratta
2 Intern
•
2K Posts
0
February 21st, 2007 22:00
safer at all. It is easier for management I would
rather say.
I agree but it goes as a practive whenevr we want to use a Storage node - direct or dedicated.
backup on storage node seems to be silently enabled
in some 7.x version. I myself have no idea if that
is ok or bug (recently I have seen variant of it
which is certainly bug), but I dislike it. Always,
and I mean always, index and bootstrap were data
which originally reside on server - thus owned by
server. Logically, they should be saved to storage
(disk or tape) assigned to server. To have storage
node save index (which is located on server) is not
something I prefer or even wish to consider unless
index db will be based on storage node in that case
(which is not). A bug I have seen is where index
gets saved under ownership of client - however that
is really under specific circumstances and
environment I doubt anyone else has.
Yes but I tried this on 6.x which I know is unsupported but worked well here.
same media as you get immediate request for tape for
pool X on server while your tape device with tape
from pool X on storage node needs to shift from
writing,idle to writing,done state. NW won't wait
for that of course (which is fine) and will load
another tape from that pool (or Default if pool X
can't be used on server) and use it.
Does this suggest I should then have a media labelled for that pool on both the server as well as storage node?
I checked in 7.3.2 admin guide that says "A bootstrap cannot be backed up to the remote device".
ble1
4 Operator
•
14.4K Posts
1
February 22nd, 2007 01:00
for that pool on both the server as well as storage
node?
Yes.
cannot be backed up to the remote device".
That is correct - as stated earlier.