You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be cool to have a way to 'import' tar archive as stdin backup.
Something like
tar c <....> | rustic backup -
Why? Sometimes it's too hard to get access to actual data that needs to be saved.
For example proxmox have vzdump command that can save backups to .tar.gz/zstd format on disk or can output tar stream to stdout. Surely we can 'unpack' that tar temporary but this extra I/O that can be avoided
I also want to add this feature but I must say that some more groundwork is needed. I'll describe a bit the problem:
rustic backup is optimized to read files in parallel (which is an already implemented performance tuning)
To allow this, the scanning of files (i.e. get the information which files are to backup) and the actual reading is separated and may occur in multiple thread
This is incompatible with reading tar data from a stream, as here we get the information about the contained file directly followed by the content before we get the information of the next file.
Besides solving this problem we also have to enhance the possibility to config backup sources, e.g. how to differentiate whether to interpret the stdin stream as tar or just as content. But this is the smaller part of the problem...
Hi,
It would be cool to have a way to 'import'
tar
archive as stdin backup.Something like
Why? Sometimes it's too hard to get access to actual data that needs to be saved.
For example
proxmox
havevzdump
command that can save backups to.tar.gz/zstd
format on disk or can outputtar
stream to stdout. Surely we can 'unpack' thattar
temporary but this extra I/O that can be avoidedSame as restic/restic#2226
The text was updated successfully, but these errors were encountered: