File constantly fails to copy from svfs

[rmufr]

Context

• svfs version : 0.9.1
• storage provider : Ovh / Hubic
• product : 10T / annual
• mount command = mount -t svfs -o block_size=4096,hubic_auth=XXX,hubic_token=YYYY,container=recovery,segment_size=4096 hubic /mnt/hubic/recovery

A file copied successfully from my local fs to svfs mountpoint with 'cp' command (return code ok). When trying to copy it back, with a 'cp' command, it fails constantly with the "unable to open file for reading : I/O error". All other files in the same directory (similar name & size) could be copied back to the local filesystem successfully. When long listing the directory, nothing seems erroneous.

Here's the debug log when listing the directory

DEBU[2017-02-20T00:09:46+01:00] <- Lookup [ID=0x86 Node=0xd Uid=0 Gid=0 Pid=1604] "sysvar_20170205T013101.041.pgp"  source=fuse                                                                                                                 
DEBU[2017-02-20T00:09:46+01:00] -> [ID=0x86] Lookup 0x38 gen=0 valid=1m0s attr={valid=1m0s ino=7689054393244243976 size=1057002091 mode=-rwx------}  source=fuse                                                                                

And when trying to copy it back from Hubic

DEBU[2017-02-20T00:09:46+01:00] <- Read [ID=0x95 Node=0xd Uid=0 Gid=0 Pid=1604] 0x1 4096 @0xca8 dir=true fl=0 lock=0 ffl=OpenReadOnly+OpenDirectory+OpenNonblock+0x20000  source=fuse
DEBU[2017-02-20T00:09:46+01:00] -> [ID=0x95] Read 0                           source=fuse
DEBU[2017-02-20T00:09:46+01:00] <- Release [ID=0x96 Node=0xd Uid=0 Gid=0 Pid=0] 0x1 fl=OpenReadOnly+OpenDirectory+OpenNonblock+0x20000 rfl=0 owner=0x0  source=fuse
DEBU[2017-02-20T00:09:46+01:00] -> [ID=0x96] Release                          source=fuse
DEBU[2017-02-20T00:10:17+01:00] <- Getattr [ID=0x97 Node=0xd Uid=0 Gid=0 Pid=1608] 0x0 fl=0  source=fuse
DEBU[2017-02-20T00:10:17+01:00] -> [ID=0x97] Getattr valid=1m0s ino=16426297346771216549 size=4096 mode=drwx------  source=fuse
DEBU[2017-02-20T00:10:17+01:00] <- Open [ID=0x98 Node=0x38 Uid=0 Gid=0 Pid=1608] dir=false fl=OpenReadOnly+0x20000  source=fuse
DEBU[2017-02-20T00:10:18+01:00] -> [ID=0x98] Open error=EIO: Object Not Found  source=fuse

Thanks in advance for your support,

RM

[rmufr]

Hi, Any clue about this issue ? Unrecoverable files is quite an anoying one not only for me. Thanks again for your efforts. Best regards, RM

[paschembri]

+1

[spirann]

try without specifying the container

[rmufr]

Sorry for giving news so late after my first post. I managed to recover the file using the following command: head -c <size_bigger_than_the_file> /path/to/file > ./recovered_file

the copy commands still failed but at least I was able to recover my file. This led me to reinforce my backup policies in the following way :

• add more redundancy (par2) files to prevent the possible loss of 1 over 10 data files and the possible loss of 1 over 10 redundancy file
• test each backup by getting back files the day after in order to validate the backup as a valid incremental backup step.