Best way to remove old backup files?

I've read through posts on this board regarding removing old .tib files so your HD doesn't fill up. I run a differential backup every night, but suppose I only want to keep the last few days worth of backups or my external HD will get close to filling up.

I have created a batch file to remove all but the first backup (ie. the full backup file) and the last 3 backup files.

Suppose I have backup.tib, backup2.tib, backup3.tib, backup4.tib, backup5.tib.

Tonight's backup runs, the pre-command runs and removes the oldest backup while preserving the original full backup. ie. "backup2.tib" is deleted.

Now the backup runs, ATI automatically names it backup2.tib since this is the first numerical filename available. It means that now my backups aren't in alphabetical order, but I can always order them my date created in explorer, so no probs.

So. All great. Except if I want to restore from a backup - by opening any of the differential backups I am presented with the list of original date/times to choose from in ATI. Although backup2.tib is my newest backup, ATI tells me it's timestamp is still the same as the ORIGINAL backup2.tib - the one that was deleted.

If I restored from backup2, can I expect any problems? Is there a tidier way of doing this, or making Acronis report the new timestamp for backup2?

 

Hi Prolecto,

In an incremental set you may delete files from the tail only. To restore from any incremental file (date) in the chain, you need all the preceeding incrementals and the full. The same holds true for validating.

By deleting the earliest incrementals you made the remaining incrementals unusable. You should now delete all the incrementals down to the full and start anew.

 

These are differential backups

 

I'm sorry, Prolecto, it was the number of differentials you have that made me interpret wrongly.

Well, with differentials you need any single differential and the full to restore. The other differentials may be present or not. So, as long as you can identify them, you are OK.

But there was a bug in TI preventing to verify the last differential image if all the previous differentials weren't present. I havent read about this bug being crushed yet.

I think that for the strategy you want to implement you would be better off with a series of full images. They will be independent, clearly identifiable and able to validate singly without problems. I also do the deleting you want to (though manually) and I always create full images only. If the bug is still there, you won't be saving any space either, since verification should certainly be performed.

 

Hello Prolecto,

Please accept our apologies for the delay with the response.

Please note that this issue was forwarded to Acronis Quality Assurance Team. I will inform you about the results. As this can take a few days, we apologize in advance for any delay with response.

Thank you.
--
Aleksandr Isakov

 

Hello Aleksandr,

While the Acronis QA Team is looking at this issue I would be grateful if you could forward the following comment to them.

Diffential Backups

A differential backup takes a backup of all changes since the Full Backup. Therefore logically a differential backup series should consist of the Full Backup and 1 (one) differential backup. (A user may for added assurance retain several differential backups, but this is a personal choice and not a technical requirement).

There is a bug in TI9 that requires all differentials to be present since the Full Backup in order to verify the last differential. This makes total nonsense of the entire logical concept and framework of differential backups. At any time in a differential backup strategy only, and I repeat ONLY, the Full Backup and 1 (one) differential need to be present for both verification and restore.

Am I annoyed? Yes. And so are a lot of other users because this bug has been around for ages and we have not received any indication that the Acronis Development Team intends fixing it.

 

Well, I don't understand this at all. What you are describing sounds exactly like a diff backup, and, as it turns out, for that you only need the full and the diff file to be able to restore with ATI.

Incs are just changes since the last inc (or full if there are no incs). So it makes sense that you would need any iterim incs to get from full to the subject inc.

Am I missunderstanding your meaning.?

 

Hi Shieber,

Yes, you are misunderstanding. In my post I wrote...

You need every differential to be present in order to verify. Acronis have agreed that it is a bug, they just don't fix it

 

Well, this sure adds a new, and concerning, wrinkle to the puzzle. I thought that I would switch over to using differential backups, rather than incremental backups, so that I did not need a chain of backups present to verify or restore.

If this is true, and I trust that it is, then apparently there is also no way to choose to restore from a differential backup that is made earlier than the latest one in the chain?

It seems to me that there should be a way, using True Image, to restore to a known reliable point in time, should a problem such as a virus infection occur. Short of having full backups (or backups and complete differential/incremental chains) on hand created at many such points, are there other ways of accomplishing this?

Why are differential backups linked? Seems to defeat their purpose, and certainly compromises the integrity of our backup system.

If such bugs exist, PLEASE tell us, so that we can protect our data.

 

Christopher_NC,

Not quite. You can Restore from any differential in a chain, but you cannot Verify your backup unless all the differentials are present.

Should you want to Restore a differential you will be given a list of all differentials and you can then choose the "last known good" differential.

But not being able to verify unless all differentials are present is just nonsense. Differential backups are not supposed to even be aware of each other. Each differential should be aware only of the last Full backup.

It is a bug and it should be fixed.

 

Hello all,

We are sorry for the delayed response.

I have contacted Acronis Development Team and would like to note that in the current version of Acronis True Image differential backup approach works the following way:

- Differential backup can be based on either FULL or any of existing differential backup archives;

- In order to restore the date from the differential backup one you must have the initial full backup and the differential backup;

- If you created several differential or both incremental and differential backups based on the same full backup (and therefore, in the same folder), you will need ALL of these successive backups to validate the archive.

This approach most likely will be changed in the further build/version of Acronis True Image. The Acronis Development Team already aware of all the suggestions made on this Forum and already took it into consideration.

Thank you.
--
Aleksandr Isakov

 

Hello Aleksandr,

Please do explain: If I create a differential based on a previous differential, will the second differential contain the changes since the full or since the differential it was based upon?

If the latter is true, would I not need the base differential as well to restore from the second differential?

 

Interesting question. Can you create a differential using another differential as the "Full" ? That is weird if true. One assumes that the original "Full" is flagged as the original in the .tib file info table and again that each differential is flagged as a differential in the .tib file info table. If you tried to create a differential using another differential as the "Full" one would have assumed that TI would throw an error.

 

I'm afraid to say that, since differential imaging was introduced into True Image, Acronis Support have provided almost as many different answers on how it's supposed to work as there have been questions . Do a forum search covering the last year, using "Acronis Support" as the user name and the keyword "Differential" and you will see what I mean. This previous thread titled <Important Questions> is a classic example.

A nice <graphic by bobat dated July 2005> clearly illustrates how I (and I believe many other users) would expect Full, Incremental and Differential images to behave and I thought Alexey Popov's reply at Post #19 in the "Important Questions" thread had set the record straight. However, Aleksandr's statement in #11 above

seems to have thrown the subject back into the melting pot .

Regards



Regards

 

Hello all,

Please note that when I said "Differential backup can be based on either FULL or any of existing differential backup archives" I meant that during differential backup creation one can select either FULL or differential backup archive. In any case the differential backup depends only on initial FULL backup. The thing is each differential backup contains information about the full backup (some kind of a "link" to the full backup), therefore, if you create the next differential backup and select the differential backup instead of the FULL one to be the "base" Acronis True Image scans the differential backup and finds this "link" to the full backup and so the differential backup contains all changes against the initial FULL archive and not against the differential. I'm sorry for not being accurate enough in my explanation.

Therefore, in order to restore the date from the differential backup one must have the initial full backup and this differential backup only. However, if you have several differential backups based on the same full backup you will need ALL of these backups to validate the archive. This issue will surely be fixed in one of the future builds/version of Acronis True Image. I'm sorry, but at the moment the exact time-frame for that is not decided yet.

The only difference between selecting the FULL or existing differential backup when creating the next differential backup is the name of the differential backup that will be generated. If you select the FULL backup during the differential backup creation the differential backup name will end with the first available number (the situation like Prolecto described in the post #1). If you select the differential backup then numeration in the differential backup names continues.

Thank you.
--
Aleksandr Isakov

 

Hello Aleksandr,

Your additional clarification is much appreciated and I'm pleased to see that there is no longer an apparent contradiction between what you and Alexey have said.

It's now perfectly clear that, when creating a new differential, it doesn't matter whether you select the full image or any one of the previous differentials; it will only ever be based on the full one.

Best regards

 

Hi Menorcaman,

You wrote...

Yes..... but remember, as Alexsandr says...

Which for an inexperienced user might be a little confusing.

Alexsandr,

While your developers are fixing the issue with having all differentials present perhaps they would consider changing the way the above works. What should happen is, if a user wants to create a differential and selects a previous differential as the base file, then TI should throw and error and popup an error message dialog box. The user should always be forced to select a Full backup as the base file for differentials. Selecting a previous incremental or differential should ALWAYS throw an error. That is the correct way to do it.

 

Hi there Tabvla,

I accept that inexperienced users may find the file numbering logic a bit confusing. However, I felt the important point was that the apparent contradiction between Alexey's and Alexsandr's statements has been cleared up. Namely, the data contained within a new differential is always based on the full image, never on a preceding differential or incremental.

Kind regards

 

Agreed, and I am glad that you pushed for a definitive answer on this issue as it has made it clearer for all of us. Tks.

 

I very much agree on that.

The option to select either file to obtain the same result doesn't make as much for additional user friendliness as it introduces confusion.

 

Perhaps Acronis based their thinking on incremental imaging, whereby you can create the next incremental by selecting either the base full image or any one of the preceding incrementals .

Regards

 

Yes, probably.

But there that solution makes sense and brings benefit to the user, since the chaining of incrementals is the normal implementation of the incremental strategy.

On the other hand, creating (to keep) several differentials based on the same full image is a waste of space in my opinion and an improper use of the differential technique (due to the repeated recording of the same changes). Therefore, I see no reason for making this procedure friendlier and promoting it. Even more so if I consider the validation bug still present.

 

I think the 3 of us are essentially in agreement on this

Backups are so critical that it really is essential that developers follow industry norms. And there is no argument that the industry norm for many years has been that incrementals chain and differentials don't. That really is the core difference between the two procedures.

Can't wait for TIv10....... OK... so I am the eternal optimist

 

I couldn't agree more. In fact the "Wink" smilie in my reply was meant to tell you that I had my tongue in cheek and didn't feel the Acronis' implementaion of differential imaging made much sense. With hind sight perhaps it would have been better if I'd used the "Dry" smilie dry instead .

Anyway, let's just hope they fix the validation bug PDQ because that definitely doesn't make sense!!

Kind regards

 

Interesting thread. It is the concept of differential backups and the ability to delete older ones in the middle of the chain that has recently made me consider upgrading from v8 but after reading this thread I think I'll hold off a while.