> On Feb 10, 2026, at 9:22 AM, Bruno Friedmann (bruno-at-bareos) 
> <[email protected]> wrote:
> 
> Hi Brock,
> 
> We have some questions here:
> 
> - Do you know the first version of Bareos which was used to create the first 
> backup ?

I don’t know but it’s probably at least 2 years old

> 
> - What is the version used for restore ?

The machine running the director, sd, and the fd I’m restoring do are current.  
I even updated them and tried again with the same results.
The first problem was noticed with 25.0.2~pre13.add249aa9 
Ran agian with: 25.0.2~pre101.a49ae5c10

> 
> - Was compression in use in the fileset, did that one changes over the year 
> (like changing from gzip to lz4 )

Compression has always been used (it’s an AWS to on prem backup so I want to 
minimize backup charges).
I did mess with compression settings a while back but I don’t htink for this 
client and it’s been gzip 6

I don’t have a way to confirm that, here is the fileset in question. 

FileSet {
  Name = "mlds_home"
  Include {
   Options {
     compression=GZIP6
     signature=MD5
   }
   File = "/mnt/xvdg/home"
  }
  Exclude {     File = "/home/mysql" # maybe future mysql bin-log location
  }
} 



> 
> - Would it be possible to send us the resulting bsr of the restore ? 

Sure attached, 
> 

-- 
You received this message because you are subscribed to the Google Groups 
"bareos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/bareos-users/A95107CB-E750-4374-9C54-05996F0406F4%40mlds-networks.com.

Attachment: myth-dir.restore.26.bsr
Description: Binary data

> 
> - Not sure to decode correctly the workflow, could you just confirm the fact
>   + Restore fails only on a specific host, with partition used for the 
> restore being an old btrfs+compression

Correct 
Note the backup source was old BTFS + Compression (was zlib moved to zstd) the 
errors were found restoring to a different machine. 
This was /home

>   + Restore can also failed on the same host with ext3 partition

This was a different fileset from that same AWS host.  this was  /  (system 
paths).
This didn’t have the compression errors though and looking at it more closely I 
would ignore this one, I think this was just files changing while the backup 
was running.
I can’t test it again because I alraedy re-ran a full and lost track of which 
restore job it was as I have been testing all my backups now.

>   + Restore always succeed if done on any other target

Ah sorry, I am not restoring to the origonal host.  All of the above were 
restoring to a directory on the server.  The whole reason I found this wasn’t 
because of a needed restore, I was just avoiding AWS bandwidth charges for a 
photo library I’m moving to Imiich locally. 

> 
> In case only one host is failing, do you already do some memory checks (if 
> any possible)?
> As the client is gentoo (no official package) do you know how compilation was 
> done ?

I’m using the masked package in their repo
https://packages.gentoo.org/packages/app-backup/bareos



> 
> Regards.
> 
> On Tuesday, 10 February 2026 at 03:56:51 UTC+1 Brock Palen wrote:
> yeah I figured that out afterwards the files that were failingchceksum were 
> all 0 size on the restore. 
> The mostly match up with the files with the Compressed header size error. But 
> not always it’s even weirder so a specific example doing a restore one of the 
> problem files: 
> 
> The restored file has 0 size, but the joblog shows a restored size that’s 
> non-zero even though it still showed error. 
> 
> root@myth:/tmp/bareos-restores# stat 
> mnt/xvdg/home/mpalen/public_html/var/resizes/Bolivia/bolivia/impedidos/07-22-08\
>  064.jpg 
> File: 
> mnt/xvdg/home/mpalen/public_html/var/resizes/Bolivia/bolivia/impedidos/07-22-08
>  064.jpg 
> Size: 0 Blocks: 0 IO Block: 4096 regular empty file 
> Device: 252,5 Inode: 26611710 Links: 1 
> Access: (0644/-rw-r--r--) Uid: (55637/ UNKNOWN) Gid: (55637/ UNKNOWN) 
> Access: 2023-07-02 02:09:03.000000000 -0400 
> Modify: 2020-02-01 17:57:47.000000000 -0500 
> Change: 2026-02-09 12:44:15.585422074 -0500 
> Birth: 2026-02-09 12:44:15.585422074 -0500 
> 
> 
> 
> The job log has many lines like and then says it restored a size different 
> than the origonal, but that restored size is non-zero. 
> 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61266, message_length=14685 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=64376, message_length=65498 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=64468, message_length=65590 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=52209, message_length=53116 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60520, message_length=61642 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60938, message_length=62060 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61158, message_length=62242 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61213, message_length=62335 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61154, message_length=62276 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61202, message_length=62324 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61305, message_length=62422 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60837, message_length=61959 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60723, message_length=61845 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60710, message_length=61832 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60754, message_length=61876 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60667, message_length=61720 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61309, message_length=62431 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61191, message_length=62313 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61352, message_length=62474 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61139, message_length=62261 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=63626, message_length=64748 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=64058, message_length=44892 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61104, message_length=62226 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61072, message_length=62194 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60934, message_length=62056 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60488, message_length=61607 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=61079, message_length=62201 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: Compressed header size 
> error. comp_len=60926, message_length=62048 
> 2026-02-09 12:44:14 myth-fd JobId 115093: Error: findlib/attribs.cc:379 File 
> size of restored file 
> /tmp/bareos-restores/mnt/xvdg/home/mpalen/public_html/var/albums/Bolivia/bolivia/impedidos/07-22-08
>  060.jpg not correct. Original 125535, restored 65536. 
> 
> 
> One thing to note this is an older btrfs filesystem with compression turned 
> on. All the other test restores (for which I have a large enough free space 
> to test) restored fine. 
> With the exception of another fileset from this same AWS host but that 
> filesystem is ext3. It was much smaller, I did a new full backup and tested 
> restore and it was fine. 
> 
> So I’m curious why this could happen, I could chalk it up to a fluke. I would 
> like to have a way to test restores for larger restores for whcih I don’t 
> have TBytes of idle disk around just in case. 
> 
> Given this has been a long running always incremental system where there any 
> bugged versions in the last few years? The client is gentoo, all my ohter 
> tests were windows, or Ubuntu. 
> 
> 
> Brock Palen 
> [email protected] 
> www.mlds-networks.com 
> Websites, Linux, Hosting, Joomla, Consulting 
> 
> 
> 
> > On Feb 9, 2026, at 10:55 AM, Bruno Friedmann (bruno-at-bareos) 
> > <[email protected]> wrote: 
> > 
> > Hi Brock, could you check what sha1sum is recorded in the database 
> > especially for those file. 
> > 
> > Seems that "da39a3ee5e6b4b0d3255bfef95601890afd80709" hash correspond to 
> > empty files. 
> > 
> > 
> > On Saturday, 7 February 2026 at 21:14:22 UTC+1 Brock Palen wrote: 
> > Oh and not all files are at issue, I ran sha1sum on the soruce host and the 
> > restored data about half the files don’t match. 
> > 
> > 
> > Brock Palen 
> > [email protected] 
> > www.mlds-networks.com 
> > Websites, Linux, Hosting, Joomla, Consulting 
> > 
> > 
> > 
> > > On Feb 7, 2026, at 3:10 PM, 'Brock Palen' via bareos-users 
> > > <[email protected]> wrote: 
> > > 
> > > I did a restore test on a host that has been using AI jobs for a while so 
> > > a lot of long lived files have been coppied form media to media for a 
> > > while, but in the restore I got a bunch of errors and the many retored 
> > > files have 0 size. 
> > > 
> > > On the host: 
> > > # ls -l P1020370.JPG 
> > > -rw-r--r-- 1 mlds-www mlds-www 2735024 Feb 1 2020 P1020370.JPG 
> > > 
> > > Restore 
> > > # ls -l P1020370.JPG 
> > > ---------- 1 root wheel 0 Feb 6 15:24 P1020370.JPG 
> > > 
> > > 
> > > 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=61092, message_length=62214 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=61039, message_length=62161 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=64376, message_length=65498 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=64549, message_length=65671 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=61025, message_length=62147 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=60966, message_length=62088 
> > > 06-Feb 15:46 myth-fd JobId 114962: Error: Compressed header size error. 
> > > comp_len=63805, message_length=64927 
> > > 
> > > 
> > > Rather than just forcing a new full backup is there a way to check my 
> > > volumes? 
> > > Or a way to do it as part of a job? 
> > > 
> > > (and this is why you test your backups). 
> > > 
> > > 
> > > Brock Palen 
> > > [email protected] 
> > > www.mlds-networks.com 
> > > Websites, Linux, Hosting, Joomla, Consulting 
> > > 
> > > 
> > > 
> > > -- 
> > > You received this message because you are subscribed to the Google Groups 
> > > "bareos-users" group. 
> > > To unsubscribe from this group and stop receiving emails from it, send an 
> > > email to [email protected]. 
> > > To view this discussion visit 
> > > https://groups.google.com/d/msgid/bareos-users/C6F19DC3-8E89-4227-9F8F-2D2AEB62DF32%40mlds-networks.com.
> > >  
> > 
> > 
> > -- 
> > You received this message because you are subscribed to the Google Groups 
> > "bareos-users" group. 
> > To unsubscribe from this group and stop receiving emails from it, send an 
> > email to [email protected]. 
> > To view this discussion visit 
> > https://groups.google.com/d/msgid/bareos-users/faf07f25-4490-4ef3-bca7-1ff8616cd125n%40googlegroups.com.
> >  
> 
> 
> -- 
> You received this message because you are subscribed to the Google Groups 
> "bareos-users" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected].
> To view this discussion visit 
> https://groups.google.com/d/msgid/bareos-users/28fa577e-dc2d-47ce-8a19-3b8a0bddf10an%40googlegroups.com.

-- 
You received this message because you are subscribed to the Google Groups 
"bareos-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/bareos-users/A95107CB-E750-4374-9C54-05996F0406F4%40mlds-networks.com.

Reply via email to