Backup Central

All-

We're trying to set up multiple streams from a single large volume on one
of our clients, using the trick with having two client instances -- one
using Saveset "All" but with a server-side directive that skips certain
directories, and the other client instance that just backs up those
particular directories that are skipped via the All. This method has
been discussed on the list before, for example see my post and Itzik
Meirson's follow-up from April 6th of 2004.

We have our special (server-side) directives set up (this is a NetWare
client):

<< "ZW01VOL1:APPS" >>
null: *
<< "ZW01VOL1:FACILITIES" >>
null: *

We have one instance of the client defined, with Saveset All and those
directives being used.

We're trying to create a second instance of the client, with Saveset set
to just

ZW01VOL1:APPS
ZW01VOL1:FACILITIES

As soon as I try put this second client in the same group as the first
one, I get an error from NetWorker:

save set of All for client foo.bar in group GroupName excludes
any clients with the same name from the group.


I can see why NetWorker would want to at least warn about this -- if
you're using "All" with a client, in theory you're backing up everything
so having a second instance of the client in the same group seems like
a waste. In practice, though, the directive controls whether anything
is really getting backed up by that client instance. In our case, there
would be no significant overlap between the two clients, but NetWorker
isn't smart enough to realize this.

Is this just a limitation at version 6, or does this still happen at v7
too?

For those of you that are using this trick to get multiple savestreams off
a big volume, have you found any way to have both instances of a client
in the same group, or are you forced to use a second group and run it at
the same time?

Tim
Tim Mooney mooney < at > dogbert.cc.ndsu.NoDak.edu
Information Technology Services (701) 231-1076 (Voice)
Room 242-J6, IACC Building (701) 231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

Hi Tim,

We're trying to set up multiple streams from a single large volume on one
of our clients, using the trick with having two client instances -- one
using Saveset "All" but with a server-side directive that skips certain
directories, and the other client instance that just backs up those
particular directories that are skipped via the All. This method has
been discussed on the list before, for example see my post and Itzik
Meirson's follow-up from April 6th of 2004.

<snip>

We have one instance of the client defined, with Saveset All and those
directives being used.

We're trying to create a second instance of the client, with Saveset set
to just

ZW01VOL1:APPS
ZW01VOL1:FACILITIES

As soon as I try put this second client in the same group as the first
one, I get an error from NetWorker:

save set of All for client foo.bar in group GroupName excludes
any clients with the same name from the group.

I can see why NetWorker would want to at least warn about this -- if
you're using "All" with a client, in theory you're backing up everything
so having a second instance of the client in the same group seems like
a waste. In practice, though, the directive controls whether anything
is really getting backed up by that client instance. In our case, there
would be no significant overlap between the two clients, but NetWorker
isn't smart enough to realize this.

Is this just a limitation at version 6, or does this still happen at v7
too?

No, it remains a limitation in NetWorker v7 as well. In fact,
NetWorker is so "smart" at detecting this that it will circumvent you
even if you force a creation of a second client in the same group
where the first client has a saveset of all:

root < at > sol
~$ savegrp -pv -c mac Default
mac:/special level=incr
mac:All level=incr
12/30/04 13:12:13 savegrp: Client mac has invalid save sets: 'All' and others
12/30/04 13:12:13 savegrp: Ignoring other save sets for mac
12/30/04 13:12:13 savegrp: Run up to 4 clients in parallel
12/30/04 13:12:13 savegrp: mac:probe started
savefs -s sol -c mac -g Default -p -l full -R -v

While I can appreciate the rationale behind the "no other client in a
savegroup when there's one with 'All' in it" rationale, I agree that
it's somewhat limiting and ends up creating situations whereby a lot
of people end up creating "inclusive" backup policies where they
explicitly include all the savesets to backup, which is often seen as
easier than maintaining two separate savegroups.

For those of you that are using this trick to get multiple savestreams off
a big volume, have you found any way to have both instances of a client
in the same group, or are you forced to use a second group and run it at
the same time?

See above - unless someone knows of a way of munging savegrp NetWorker
does it's best to force you to not do this, even if you circumvent it
:(

Cheers,

-Preston de Guise.

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

This seems like a lot of work to get multiple save streams going. Whats
wrong with breaking the client backup into 3-4 save streams by specifying a
portion of the backup job instead of using all?

IE insted of all on a Windows PC, you specify
c:\
d:\
e:\

if the PC has 3 disks. If it has a single large disk, specify
c:\Directory1
c:\directory2
c:\Directory3

Each of these will start 3 save streams going and that's one more than you
get with your method. The only problem with this is if you have a single
directory with 30-40 hugh files (I had this in an Oracle database backup.)
and you want multiple save streams. That was a problem because it was a
UNIX client and I couldn't use wildcards.


However, with that exception, I have always been able to tune the backups to
run however I wanted with this method.

Just a thought

Calvin Thomas
UNIX System Administrator
NACA Logistics IT



-----Original Message-----
From: Legato NetWorker discussion
[mailto:NETWORKER < at > LISTMAIL.TEMPLE.EDU]On Behalf Of Preston de Guise
Sent: Wednesday, 29 December, 2004 06:28 PM
To: NETWORKER < at > LISTMAIL.TEMPLE.EDU
Subject: Re: [Networker] multiple instances of a client in one group?


Hi Tim,

We're trying to set up multiple streams from a single large volume on one
of our clients, using the trick with having two client instances -- one
using Saveset "All" but with a server-side directive that skips certain
directories, and the other client instance that just backs up those
particular directories that are skipped via the All. This method has
been discussed on the list before, for example see my post and Itzik
Meirson's follow-up from April 6th of 2004.

<snip>

We have one instance of the client defined, with Saveset All and those
directives being used.

We're trying to create a second instance of the client, with Saveset set
to just

ZW01VOL1:APPS
ZW01VOL1:FACILITIES

As soon as I try put this second client in the same group as the first
one, I get an error from NetWorker:

save set of All for client foo.bar in group GroupName excludes
any clients with the same name from the group.

I can see why NetWorker would want to at least warn about this -- if
you're using "All" with a client, in theory you're backing up everything
so having a second instance of the client in the same group seems like
a waste. In practice, though, the directive controls whether anything
is really getting backed up by that client instance. In our case, there
would be no significant overlap between the two clients, but NetWorker
isn't smart enough to realize this.

Is this just a limitation at version 6, or does this still happen at v7
too?

No, it remains a limitation in NetWorker v7 as well. In fact,
NetWorker is so "smart" at detecting this that it will circumvent you
even if you force a creation of a second client in the same group
where the first client has a saveset of all:

root < at > sol
~$ savegrp -pv -c mac Default
mac:/special level=incr
mac:All level=incr
12/30/04 13:12:13 savegrp: Client mac has invalid save sets: 'All' and
others
12/30/04 13:12:13 savegrp: Ignoring other save sets for mac
12/30/04 13:12:13 savegrp: Run up to 4 clients in parallel
12/30/04 13:12:13 savegrp: mac:probe
started
savefs -s sol -c mac -g Default -p -l full -R -v

While I can appreciate the rationale behind the "no other client in a
savegroup when there's one with 'All' in it" rationale, I agree that
it's somewhat limiting and ends up creating situations whereby a lot
of people end up creating "inclusive" backup policies where they
explicitly include all the savesets to backup, which is often seen as
easier than maintaining two separate savegroups.

For those of you that are using this trick to get multiple savestreams off
a big volume, have you found any way to have both instances of a client
in the same group, or are you forced to use a second group and run it at
the same time?

See above - unless someone knows of a way of munging savegrp NetWorker
does it's best to force you to not do this, even if you circumvent it
:(

Cheers,

-Preston de Guise.

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

The major problem with this is that unless you are a small site you need a
rock solid process to modify the client definitions when disks are added
or with sub-directories directories are added or renamed. This is managed
better with either a script that checks and modifies the client savesets
to match the current reality, or as suggested but with the all client in a
different savegroup - we use the latter for a big backup where the fulls
are spread out over several nights but any changes are caught by the all
groups which uses a custom directive to ignore the partitions backed up by
the other client savesets.


On Thu, 30 Dec 2004 04:03:30 -0800, Calvin Thomas
<calvin.thomas < at > NACALOGISTICS.COM> wrote:

This seems like a lot of work to get multiple save streams going. Whats
wrong with breaking the client backup into 3-4 save streams by
specifying a
portion of the backup job instead of using all?

IE insted of all on a Windows PC, you specify
c:\
d:\
e:\

if the PC has 3 disks. If it has a single large disk, specify
c:\Directory1
c:\directory2
c:\Directory3

Each of these will start 3 save streams going and that's one more than you
get with your method. The only problem with this is if you have a single
directory with 30-40 hugh files (I had this in an Oracle database backup.)
and you want multiple save streams. That was a problem because it was a
UNIX client and I couldn't use wildcards.


However, with that exception, I have always been able to tune the backups
to
run however I wanted with this method.

Just a thought

Calvin Thomas
UNIX System Administrator
NACA Logistics IT



-----Original Message-----
From: Legato NetWorker discussion
[mailto:NETWORKER < at > LISTMAIL.TEMPLE.EDU]On Behalf Of Preston de Guise
Sent: Wednesday, 29 December, 2004 06:28 PM
To: NETWORKER < at > LISTMAIL.TEMPLE.EDU
Subject: Re: [Networker] multiple instances of a client in one group?


Hi Tim,

We're trying to set up multiple streams from a single large volume on
one
of our clients, using the trick with having two client instances -- one
using Saveset "All" but with a server-side directive that skips certain
directories, and the other client instance that just backs up those
particular directories that are skipped via the All. This method has
been discussed on the list before, for example see my post and Itzik
Meirson's follow-up from April 6th of 2004.

<snip>

We have one instance of the client defined, with Saveset All and those
directives being used.

We're trying to create a second instance of the client, with Saveset set
to just

ZW01VOL1:APPS
ZW01VOL1:FACILITIES

As soon as I try put this second client in the same group as the first
one, I get an error from NetWorker:

save set of All for client foo.bar in group GroupName excludes
any clients with the same name from the group.

I can see why NetWorker would want to at least warn about this -- if
you're using "All" with a client, in theory you're backing up everything
so having a second instance of the client in the same group seems like
a waste. In practice, though, the directive controls whether anything
is really getting backed up by that client instance. In our case, there
would be no significant overlap between the two clients, but NetWorker
isn't smart enough to realize this.

Is this just a limitation at version 6, or does this still happen at v7
too?

No, it remains a limitation in NetWorker v7 as well. In fact,
NetWorker is so "smart" at detecting this that it will circumvent you
even if you force a creation of a second client in the same group
where the first client has a saveset of all:

root < at > sol
~$ savegrp -pv -c mac Default
mac:/special level=incr
mac:All level=incr
12/30/04 13:12:13 savegrp: Client mac has invalid save sets: 'All' and
others
12/30/04 13:12:13 savegrp: Ignoring other save sets for mac
12/30/04 13:12:13 savegrp: Run up to 4 clients in parallel
12/30/04 13:12:13 savegrp: mac:probe
started
savefs -s sol -c mac -g Default -p -l full -R -v

While I can appreciate the rationale behind the "no other client in a
savegroup when there's one with 'All' in it" rationale, I agree that
it's somewhat limiting and ends up creating situations whereby a lot
of people end up creating "inclusive" backup policies where they
explicitly include all the savesets to backup, which is often seen as
easier than maintaining two separate savegroups.

For those of you that are using this trick to get multiple savestreams
off
a big volume, have you found any way to have both instances of a client
in the same group, or are you forced to use a second group and run it at
the same time?

See above - unless someone knows of a way of munging savegrp NetWorker
does it's best to force you to not do this, even if you circumvent it
:(

Cheers,

-Preston de Guise.

--
Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

--
Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu

In regard to: Re: [Networker] multiple instances of a client in one group?,...:

This seems like a lot of work to get multiple save streams going.

It is, but since I only have administrator (or even login) access on
half of the systems we back up, it's the only way I can guarantee that
everything is getting backed up.

Using your method, if the administrator for the box adds a new directory
or a new volume and forgets to tell me, we potentially wouldn't discover
it until the worst possible time.

Whats
wrong with breaking the client backup into 3-4 save streams by specifying a
portion of the backup job instead of using all?

Nothing, if you can guarantee that you never miss any new directories or
volumes. I can't make that guarantee.

Tim
Tim Mooney mooney < at > dogbert.cc.ndsu.NoDak.edu
Information Technology Services (701) 231-1076 (Voice)
Room 242-J6, IACC Building (701) 231-8541 (Fax)
North Dakota State University, Fargo, ND 58105-5164

Note: To sign off this list, send a "signoff networker" command via email
should be sent to stan < at > temple.edu