Backup Central

I thought for sure I heard that NW 7.6.2 could do this. I have storage
nodes that have AFTD devices. And due to some SAN re-configuration, these
devices will go from being larger than 2TB to being limited to 2TB (it's
complicated).

Anyway, what I need to do is then present 2 of these 2TB devices to my
snode, and have it so that when the first one fills up, it will
automatically "spill over" and start to use the next one. In effect, I
need the savesets to span across multiple AFTD devices. I don't want to
use both at once; I want it to keep writing to AFTD device #1 until it
runs out of space, then automatically continue writing to AFTD device #2.
(I'm not looking to "load balance" or write to both AFTD devices at the
same time)

But I am unsure how to accomplish that, and my searching skills are
failing me this morning.

I've found messages stating that you can (or "will be able to"; it's an
old posting).
<https://community.emc.com/message/466482 >

Can anyone help? Pointers, links, etc?

--
Michael Leone
Network Administrator, ISM
Philadelphia Housing Authority
2500 Jackson St
Philadelphia, PA 19145
Tel: 215-684-4180
Cell: 215-252-0143
<mailto:michael.leone < at > pha.phila.gov>


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

Hi Michael,

an AFTD media will never be set to full AUTOMATICALLY. So this will not work.
I am pretty sure that you might set it to full but not when there is a pending write. Consequently, if you abort the write process, the aborted save set will automatically be deleted which means that a save set can never be continued on another media (no matter which type).

You can achieve what you want but in this case you must use the predecessor, the FTD device type.

an AFTD media will never be set to full AUTOMATICALLY. So this will not
work.
I am pretty sure that you might set it to full but not when there is
a pending write. Consequently, if you abort the write process, the
aborted save set will automatically be deleted which means that a
save set can never be continued on another media (no matter which type).

You can achieve what you want but in this case you must use the
predecessor, the FTD device type.


Huh. I completely forgot about FTD devices, since I always use AFTD ..

So here's what I want:

Write to device.
Automatically clone to tape when backup finished. (this is what my AFTD
devices do now)

I want the data to stay on the device until I manually clear it with a
script (this way, I have multiple days backup available immediately on
disk, for quick recovers).
So if the FTD device gets full, I want it to just sit, until I manually
clear it with a script as a scheduled task. (I parse an mminfo command,
and then remove the savesets I've already cloned that are of a certain age
with nsrmm, and clear AFTD space with "nsrstage -C -V".

I assume I can (must?) do the same with the FTD device?

I'm still unclear how NW will keep using a device until full (I get that I
must somehow set the capacity of the FTD device to be the size of the
LUN), and then automatically starts using the next device. And when I
clear space on Device #1 (with that scheduled script), will NW go back and
re-use that space, even if Device #2 has space available?


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

On 24/04/12 16:20, Michael Leone wrote:
I thought for sure I heard that NW 7.6.2 could do this. I have storage
nodes that have AFTD devices. And due to some SAN re-configuration, these
devices will go from being larger than 2TB to being limited to 2TB (it's
complicated).

Anyway, what I need to do is then present 2 of these 2TB devices to my
snode, and have it so that when the first one fills up, it will
automatically "spill over" and start to use the next one. In effect, I
need the savesets to span across multiple AFTD devices. I don't want to
use both at once; I want it to keep writing to AFTD device #1 until it
runs out of space, then automatically continue writing to AFTD device #2.
(I'm not looking to "load balance" or write to both AFTD devices at the
same time)

But I am unsure how to accomplish that, and my searching skills are
failing me this morning.

AFTD devices still do NOT span at any released version. The only thing
you could do is set a staging policy to stage between devices when one
fills, but this is a bit inefficient.


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

FTDs just behave like tapes. The main differences to their predecessor, the 'optical' device media are
- FTD media can handle individual save sets
- FTD media can only read OR write

Besides that, the FTD media will only be overwritten when
- All save sets on the media have expired (are in the state 'recyclable') so the media became 'recyclable' too AND
- NW needs a media from the same pool on this SN OR
- The administrator relabels the media

Can you leverage your OS to re-assemble the two LUNs into one 4 TB device?

Solaris - SVM/DiskSuite or ZFS
Windows - Dynamic Disk
Linux - LVM volume

--TSK

-----=====
Tim Kimball
http://sungak.net
=====-----


-----Original Message-----
From: Michael Leone [mailto:Michael.Leone < at > PHA.PHILA.GOV]
Sent: Tuesday, April 24, 2012 11:21 AM
Subject: Spanning savesets across multiple AFTD devices

I thought for sure I heard that NW 7.6.2 could do this. I have storage
nodes that have AFTD devices. And due to some SAN re-configuration, these
devices will go from being larger than 2TB to being limited to 2TB (it's
complicated).

Anyway, what I need to do is then present 2 of these 2TB devices to my
snode, and have it so that when the first one fills up, it will
automatically "spill over" and start to use the next one. In effect, I
need the savesets to span across multiple AFTD devices. I don't want to
use both at once; I want it to keep writing to AFTD device #1 until it
runs out of space, then automatically continue writing to AFTD device #2.
(I'm not looking to "load balance" or write to both AFTD devices at the
same time)

But I am unsure how to accomplish that, and my searching skills are
failing me this morning.

I've found messages stating that you can (or "will be able to"; it's an
old posting).
<https://community.emc.com/message/466482 >

Can anyone help? Pointers, links, etc?

--
Michael Leone
Network Administrator, ISM
Philadelphia Housing Authority
2500 Jackson St
Philadelphia, PA 19145
Tel: 215-684-4180
Cell: 215-252-0143
<mailto:michael.leone < at > pha.phila.gov>


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

Can you leverage your OS to re-assemble the two LUNs into one 4 TB
device?

Solaris - SVM/DiskSuite or ZFS
Windows - Dynamic Disk
Linux - LVM volume

Can I? Yes. Will I? Not if I don't have to ...

If there is a problem with one LUN, the whole disk becomes inaccessible
(at least under Windows, probably with the other Oses). With 2 separate
LUNs, it is still possible to write to/recover from the non-problem LUN.

"Dynamic Disk" is a lot like RAID-0, which I am always against, except for
things like scratch disks.


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

FTDs just behave like tapes. The main differences to their
predecessor, the 'optical' device media are
- FTD media can handle individual save sets
- FTD media can only read OR write

Shouldn't be a problem, as I won't be reading when I'm writing ..

Besides that, the FTD media will only be overwritten when
- All save sets on the media have expired (are in the state
'recyclable') so the media became 'recyclable' too AND
- NW needs a media from the same pool on this SN OR
- The administrator relabels the media

I spoke with EMC Tech Support. I explained that I manually remove savesets
that have been cloned, after a certain number of days (using "nsrmm -d
-S"), and then reclaim storage using "nsrstage -C -V". I asked about doing
this explicitly. That if Device #1 was full, and would savesets then
continue onto Device #2; they said yes. I said "what about when I manually
clear space on Device #1 using "nsrmm -d" and "nsrstage -C -V"; would NW
then be able to use the now free space on Device #1? They said that NW
would continue to use Device #2 until it filled up, and then would go back
and use Device #1, since it now had free space.

Are you saying that is NOT what would happen? That I would have to
manually re-label Device #1 before I can use it again? I haven't time (nor
the resources) to test that yet.

All my (A)FTD devices are in/will be in the same pool. Is that not a good
idea? It's been working fine as AFTD device for the last few years.


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

I spoke with EMC Tech Support. I explained that I manually remove savesets
that have been cloned, after a certain number of days (using "nsrmm -d
-S"), and then reclaim storage using "nsrstage -C -V".

>> This is nonsense. What do you want to stage if the save set has already been deleted?
AFAIR if you manually delete a save set it is deleted (space is released) right away.
Just take a few minutes and test it on a new device.


I asked about doing
this explicitly. That if Device #1 was full, and would savesets then
continue onto Device #2; they said yes. I said "what about when I manually
clear space on Device #1 using "nsrmm -d" and "nsrstage -C -V"; would NW
then be able to use the now free space on Device #1? They said that NW
would continue to use Device #2 until it filled up, and then would go back
and use Device #1, since it now had free space.

>> Correct with respect to the sequence, wrong for the procedure (see above).


Are you saying that is NOT what would happen? That I would have to
manually re-label Device #1 before I can use it again? I haven't time (nor
the resources) to test that yet.

>> Yes, if you want to reuse the whole space, labeling would be the fastest operation.


All my (A)FTD devices are in/will be in the same pool. Is that not a good
idea? It's been working fine as AFTD device for the last few years.

>> It is o.k. if your network/data path is fast enough to feed enough data to the devices.

On 26/04/12 15:20, bingo wrote:
I spoke with EMC Tech Support. I explained that I manually remove savesets
that have been cloned, after a certain number of days (using "nsrmm -d
-S"), and then reclaim storage using "nsrstage -C -V".

This is nonsense. What do you want to stage if the save set has already been deleted?
AFAIR if you manually delete a save set it is deleted (space is released) right away.
Just take a few minutes and test it on a new device.


Makes perfect sense to me. You don't want to stage immediately because
you want to keep it on disk for as long as possible, but you do want to
get the data off to tape as soon as possible. You then clear space using
nsrmm -d but only when that space is needed.

I wrote a suite of scripts to automate this a few years ago. They're
probably still floating around on the net somewhere.


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

I spoke with EMC Tech Support. I explained that I manually remove
savesets
that have been cloned, after a certain number of days (using "nsrmm -d
-S"), and then reclaim storage using "nsrstage -C -V".

This is nonsense. What do you want to stage if the save set has
already been deleted?

That's how the scripts were set up by EMC Tech Support when we first
installed NW. It's worked fine for our needs for over 5 years, so why
change, if the current way does what we need? The savesets are only
deleted after they have already been cloned, and a certain amount of time
has passed (anywhere from 2 days to a week, depending on how long we want
to keep a copy of the savesets around for faster recovers).

AFAIR if you manually delete a save set it is deleted (space is
released) right away.
Just take a few minutes and test it on a new device.

I don't have a spare snode to test it on. And I don't want to/can't make
such investigational changes to production storage nodes. Nor is there any
free SAN space to present as 2 new testing drives.



I asked about doing
this explicitly. That if Device #1 was full, and would savesets then
continue onto Device #2; they said yes. I said "what about when I
manually
clear space on Device #1 using "nsrmm -d" and "nsrstage -C -V"; would NW
then be able to use the now free space on Device #1? They said that NW
would continue to use Device #2 until it filled up, and then would go
back
and use Device #1, since it now had free space.

Correct with respect to the sequence, wrong for the procedure (see
above).


Are you saying that is NOT what would happen? That I would have to
manually re-label Device #1 before I can use it again? I haven't time
(nor
the resources) to test that yet.

Yes, if you want to reuse the whole space, labeling would be the
fastest operation.

OK, but I'm not asking what would be fastest; I'm asking if FTD devices
would just automatically re-use Device #1, since it now had free space.
That's what the AFTD device does, and what I need to happen. I can't have
NW just sit there when Device #2 fills up, waiting for me to come along
and manually re-label Device #1 (which will still have non-expired
savesets on it that I want to keep around for a while, and therefore the
whole volume is not recyclable).


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

On 26/04/12 15:20, bingo wrote:
I spoke with EMC Tech Support. I explained that I manually remove
savesets
that have been cloned, after a certain number of days (using "nsrmm -d
-S"), and then reclaim storage using "nsrstage -C -V".

This is nonsense. What do you want to stage if the save set has
already been deleted?
AFAIR if you manually delete a save set it is deleted (space is
released) right away.
Just take a few minutes and test it on a new device.


Makes perfect sense to me. You don't want to stage immediately because
you want to keep it on disk for as long as possible, but you do want to
get the data off to tape as soon as possible. You then clear space using
nsrmm -d but only when that space is needed.

Exactly what we do. The savesets are cloned to tape immediately, then we
keep those savesets around on disk for anywhere from 2 days to a week,
depending on how big the savesets are, how big the disk space is, and how
important it is to do immediate recovers of (recent) savesets. The tapes
get sent offsite the next day.

I wrote a suite of scripts to automate this a few years ago. They're
probably still floating around on the net somewhere.

I wouldn't mind seeing those. My scripts are Windows based, but the
logic is probably the same. I pipe a mminfo on savesets of a specific age,
that have been cloned (copies = 3; 2 on disk, 1 on tape). I use that as
the basis to delete (nsrm -d). Then reclaim disk space with "nsrstage -C
-V".


via RSS at http://listserv.temple.edu/cgi-bin/wa?RSS&L=NETWORKER

Just run the following short test:
- create 2 new pools (no selection criteria)
- create 2 new FTDs (1 as backup, 1 as stage destination) - you do not need but a few MB of space
- label the FTD media to the new pools
- manually backup a few MB to the first FTD (save -b bu_pool pathname)
- verify the FTD directory (they are easy to check)
- get the SSID
- stage the SS with "nsrstage -mv -b stage_pool"
- verify both FTD directories

It should not take you more than 15 mins. I am pretty sure you do not need anything else.

And if you do not want to disturb production, just install NW on any server or workstation (yes, it runs!) in eval mode (15 more mins).