Backup file names in U4 : added 4 extra hex symbols

[RubinCompServ]

For some reason, Engineering changed the file naming convention for the backup files. Now, instead of a backup file named based on the job, date/time, and file type (such as “SQLTRND2019-01-02T121324.vrb”), the file is named based on the job, gibberish, date/time, and file type (“SQLTRN_CE6DD2019-01-19T120918.vrb”). This makes a directory listing (or even viewing Backup Properties within VBR) more complicated, as sorting by name has become useless. Was Veeam so concerned about multiple jobs with the same name firing off at the exact same second that they had to introduce 4 random characters right in the middle of the file name?

[Dima P.]

Hello David,

Thanks for bringing this up! The file names were modified to improve the stability and avoid possible issues with per-vm backup chains functionality (to avoid possible name collisions we've added four symbols GUID to the name).

This makes a directory listing (or even viewing Backup Properties within VBR) more complicated, as sorting by name has become useless.

Backup properties dialog was reworked in last version, so you can use the sorting based on name or date, does that help in your case? Thank you!

[RubinCompServ]

Dima,

Unfortunately, sorting by name is meaningless now, because of the added four symbols in the middle. Had they been placed at the end, the filenames would have retained their meaning.

[Gostev]

I do agree this is a bad design. However, knowing some internal architecture details, I do see how it was safer not to change the ending of backup file name without potentially introducing lots of bugs.

Anyway, it is also totally unexpected change to me because for some reason it was done outside of our standard change process. I assume this was due to the fact that this change was done as a part of a bug fix. Let me see if we can move these symbols to the end of backup file name right in the next update.

Thanks!

[RubinCompServ]

Thank you Gostev!

[jb238]

With update 4 of B&R 9.5 Veeam changed the backup filenames.
Now they look like <job>_<4 hex digits><yyyy-mm-dd>T<hhmmss>.
The random hex digits are BEFORE the timestamp. This means, that alphabetically sorting the files no longer is meaningful.
It also means that our application to distribute the latest full backup of all jobs to multiple USB disk no longer works.
Does anybody know what the point of these hex digits is and why Veeam has to change the backup filenames again?

[veremin]

Your post has been merged into existing discussion.

Kindly, check the answers provided above and ask additional questions, if any are left.

Thanks!

[Gostev]

All,

After reviewing this internally, we do really need this random sequence added for backup name uniqueness, to avoid bugs due to file name collisions. While it sounds highely unlikely, with 320K+ customers this turns into quite a problem in absolute numbers.

However, we should be able to move this "salt" to the end of the backup file name:

Before: SQLTRN-D2019-01-19T120918.vrb
U4: SQLTRN_CE6DD2019-01-19T120918.vrb
U4a: SQLTRN-D2019-01-19T120918_CE6D.vrb

Does this look good to everyone?

Thanks!

[GregorD]

its better yes at the end.
when will be 4a release?

[RubinCompServ]

U4a: SQLTRN-D2019-01-19T120918_CE6D.vrb

That would be perfect for us. Thank you Gostev!

[Gostev]

It's not defined yet, as with bug fix updates we wait until there's enough known issues to justify shipping the new build. Historically, it takes 2-3 months after a major release, although there were releases where we did not need such a bug fix update at all. But I don't think Update 4 has a chance to be such a release, just based on the number of new features!

[doktornotor]

+1 on this, highly annoying.

Also, for agent-based jobs, the FQDN part of the file name is somehow being truncated now for some weird reason with no clear pattern (truncated to 13 chars on one server, 21 on another...)

[Dima P.]

doktornotor,

the FQDN part of the file name is somehow being truncated now for some weird reason with no clear pattern (truncated to 13 chars on one server, 21 on another...)

Thank you for sharing. I'll discuss it with the agent team. Cheers!

[KelKnowles]

I noticed that the current naming convention with Veeam Backup & Replication 9.5 Update 4 does not work well when attempting to sort it by date created.

The current naming convention seems to be: JobName_#FIVE DIGIT RANDOMID#YYYY-MM-DDHHMMSS.vib

I would recommend putting the #FIVE DIGIT RANDOMID# at the end of the filename

Since the date modified on files is not always the same as the date of the data inside of a .vib or .vbk I would like the ability to sort the repository by file name and have it not look like this:

[HannesK]

Hello,
thanks for bringing that up - we are looking to improve that.

Best regards,
Hannes

[veremin]

This will be addressed in U4a, the "randomfivedigit" stuff will be added to the end of the backup file name. Thanks!

[RubinCompServ]

Just as a reminder for people using Linux to look at their files, an 'ls -tr' will sort files by modified date, with the earliest first.

[ferrus]

Can anyone confirm if the fix for this bug, made it into Update 4a?
I couldn't see it in the release notes.

[cmaier]

Yes, I can confirm it.

[davide.asts]

Hi,

I have noticed that after the U4 filenames include an "hex uuid" (different for each file): for example:

SERVER1.vm-123_857ED2019-04-01T220000.vib
SERVER1.vm-123_8E80D2019-04-03T220000.vib
SERVER1.vm-123_B699D2019-04-07T220000.vib

So doing the lexicographic sort doesn't match anymore the date/time sort.

Isn't possible to add this uuid AFTER the date/time section of the filename?

[PTide]

Hi,

This was addressed in U4a, so you can fix this by updating.

Thanks

[vanbeek]

This is still very annoying, all automated scripts handling the backup files, need to be rewritten. sorting on name inside veeam is the least of the issue with this name convention change.

if veeam is concerned about duplicate filenames, why is this only a worry with the vbk files and not with the vbm files?

it was working without issues for the last 5+ years. make it an option, for the people that have trouble with filenames not been unique. or atleast make the vbk and vbm files that are the same backup job have the same name.

[Gostev]

This cannot be an option, as name collision means corrupted backup.

[yakamoneye18]

This is still very annoying, all automated scripts handling the backup files, need to be rewritten. sorting on name inside veeam is the least of the issue with this name convention change.

Hi Vanbeek,

just out of curiosity - which tasks are you handling via script? In my view, most tasks can be handled with Veeam itself.

And even without the extra random characters - with the timestamp in the filename (even with seconds, and they are also random), it is already impossible to work with fix filenames in script. I guess most scripting languages have the ability to find files with wildcards, which you would need with or without the random characters.

Regards,

Tobias

[soncscy]

Yeah, I had read Vanbeek's response earlier and not had a chance to write up a response, but Vanbeek, what exactly are you doing that you need the explicitly named file and can't just mask the files? I'm mostly just curious what you're doing.