Rotating – Three Sets or Three Members?

[alabay]

May I ask something regarding rotation ...

 

I have three external FW disks,, names A, B and C. Monday and thursday A, tuesday and friday B, wednesday, saturday and sunday C. That was my rotation principle for years (back then with DDS and ADR tapes).

 

Now I have one proactive backup script, one source, three media sets and one schedule.

 

Problem: RS8 doesn't check what disk/media set is "online" *before* it starts to scan. So sometimes it starts Backup A, although harddisk B is online, waiting. Very stupid, should be fixed.

 

Now, the Iomega support told me, instead I should use ONE media set with thre MEMBERS.

 

So, what is better for rotating backups?

 

On these disks there are also other backups from other sources, all differing as A, B or C.

 

I forgot ... schedule is every 20 hours all day. So it is not completely fixed that A has to be mondays, if needed, it can be redone tuesdays. This would be the next question ... how to schedule this best?

[Maser]

If you have all media sets in one script, I don't believe that works to rotate. (Maybe it's supposed to?)

 

Why not make three proactive scripts - one for each media set -- that will only run on specific days?

 

Meaning "Pro Script A" backs up to Media Set A on Disk A -- and runs M and Th.

 

"Pro Script B" backs up to Media Set B on Disk B -- T & F

 

"Pro Script C" backs up to Media Set C on Disk C -- W, Sat, Sun

 

 

Would this not work?

 

 

What I don't know -- as I've never tried it -- is whether or not you can hot-plug in the external FW disk and have it "seen" by the engine -- without stopping/starting the engine in between.

 

I've always stopped the engine when switching external disks.

[RobbL]

I have two sets with two scripts using two external FW drives. Media set A backs up one week and then Media set B does the next week.

 

I have had no issues with simply unmounting one drive and replacing it with the other one. I don't shut down the retrospect engine but I have not had a problem. The engine does not seem to notice whether the media is there or not until it tries to write to it..

[alabay]

Well, sometimes my principle works. Sometimes RS starts to scan without checking, which one of the HD is reachable.

 

I don't hotplug, all three disks are plugged to the machine, but only one is switched on, two are off.

 

Once I tried this with separate proactive scripts, but it did not work. Schedulling seemed not to be possible with saying "these two or three days, the whole day".

 

But in RS6 I used this technique, too - one script with three destinations. RS6 first checked the destinations, then started to scan resources.

 

I think, this is the problem.

 

But I would like to know how it is to make one script with one set with three members. Same? Better? Worse?

[kyeoh]

Agree with the OP. Worked in Retrospect 6, not in Retrospect 8. Wastes time, but seems to figure it out eventually.

 

The reason not to have multiple sets for me is to back up to whichever backup drive happens to be available at any one time. I might forget to rotate & it will still work to whatever it can find.

[binba]

I wanted it to work like that, doesn't seem to. Even after circumventing this, and creating member 2-weekly, it demands member 1-weekly.

 

The correct solution might be to create a media set per disk, but I have 7 media sets (per source) and don't feel like having 14. Frankly, after all the hours I've spent troubleshooting Retrospect problems I don't feel confident putting all my eggs in one basket -- I mean, catalog file.

[demani]

One media set will not work. One script to three sets can be made to work. Three scripts going to one set each are easiest to deal with.

 

The key is to make sure your preferences are set: The media time out should be set to something short (like 5 minutes), and the option to automatically use new media if an existing member is not found should be off (really only needed for the first round of backups, but it's better to have it set that way).

 

 

And Rob: turning off the power is the same as pulling the cable- that's a hot swap since the system can no longer see the drive.

 

When A doesn't find its media it should wait 5 minutes for media, then and try again with B, etc.

 

That does work (I've done it in many locations). I actually prefer to use three scripts though, that makes things easier to keep track of logically (I do just leave them set to run every night, as mentioned that will make sure a script runs on any given night).

Edited October 4, 2010 by Guest

[CallMeDave]

Even after ... creating member 2-weekly, it demands member 1-weekly.

 

 

Retrospect writes to Members in sequential order. When a Member has been filled, or lost, or set aside in a "Skip to new media" script option, Retrospect will request a new Member and continue to use that until another new one is again necessary.

 

That's how the program has always worked.

[demani]

Yeah, I'm coming back to this and I'm again confused how it would be possible to have a single set where the data was rotated among three members. Can anyone explain how they actually did that in 6? My understanding is that they were always serial, not isolated individual members. Multiple sets is easy to describe and figure out, but I just can't wrap my head around multiple member rotation.

[CallMeDave]

I'm again confused how it would be possible to have a single set where the data was rotated among three members.

 

It wouldn't be possible.

 

Can anyone explain how they actually did that in 6?

 

No one actually did.

 

I just can't wrap my head around multiple member rotation.

 

Probably because it doesn't work.

[demani]

That's what I thought...so I guess either the OP misheard the support technician, or the support technician was really wrong. But that technician comment really threw me for a loop.

[rhwalker]

I'm again confused how it would be possible to have a single set where the data was rotated among three members.

 

It wouldn't be possible.

No, it just isn't possible with the current design.

 

If you think about it, the media set / backup set / data set, whatever the current marketing speak name is - it keeps changing - is just a database, and the catalog is a database index into the database.

 

It would be possible, but not with the current design, to just treat the members as a bit bucket, and use whatever is available, providing that the member capacity hasn't been reached. That would solve the many-years-running request for moving disks offsite.

 

Left as an exercise for the programmers.

 

Russ

[dangermouse]

This works just fine for me. I have 5 HD's. Each is a separate Media Set. One is always offsite for several weeks at a stretch, two are online and in use and two more are in a drawer.

 

Retrospect has a proactive script with a number of sources, each backed up daily. Each source gets backed up to the available media sets, alternating between each daily.

 

Once a week I swap out one active disk for one from the drawer. Every few weeks I swap one from the drawer for the offsite one.

 

Retrospect always shows me which sets it thinks are in use and plans to use, but always, at some point after I swap a disk, it spots the change (within 10-15 minutes, maybe?) and shows it plans to use the new disk, which it later does corrrectly.

 

I used to have your problem and would have to stop the first failing run to the no-longer-available-disk, after which it would spot the change and update all other planned backups. A while ago this problem stopped after an update to a newer version. 8.2? I honestly don't remember, but either that or 8.1.

 

Not sure I've helped, but knowing it can work reliably may be useful?

[rhwalker]

I have three external FW disks,, names A, B and C.

 

Hmmmm. Then perhaps this is related to the bug whereby Retrospect barfs on single-character volume names?

 

Volumes don't show up if single-character names

 

I may have misread the original problem statement in haste - I didn't fully appreciate that these were Proactive scripts.

 

Russ

[alabay]

No, exactly they are named "Retrospect-X", X indicating A, B or C. Since the last update the recognition of an online disk is faster, it seems. Sad that it doesn't recognize it in seconds but in 10-15 minutes. But the bug, starting to collect files for a backup without having handshaked with the harddisk where all should be saved to, seems to be gone. Hopefully.