fcorbelli / zpaqfranz

Deduplicating archiver with encryption and paranoid-level tests. Swiss army knife for the serious backup and disaster recovery manager. Ransomware neutralizer. Win/Linux/Unix
MIT License
275 stars 25 forks source link

Broken help #38

Closed ratay closed 1 year ago

ratay commented 1 year ago

Hi, I found that somewhere around version 55.10 full help (h h) stopped working and seems -diff is also not available for a longer time

is it possible to consider support for some helpful functions on linux in the future ?

thanks Ratay

fcorbelli commented 1 year ago

Help is reworked to become more similar to debian package (short version)
For symlink etc yes,I can. But a lot of work is needed.

fcorbelli commented 1 year ago

Update
I am looking for tar-like metadata

ratay commented 1 year ago

good to know with metadata and symlink support it would be really great full-featured backup tool (but without it something important will always be missing) thanks for the great work

fcorbelli commented 1 year ago

If you have any suggestions (even on help) please write. Currently in development there is Windows dd-images (even of C:)

ratay commented 1 year ago

My programming skills on linux are read-only :(( So from what I read: You read file attributes via stat(...) In addition to the currently processed one, there are also other important ones: st_uid, st_gid, S_ISLNK, ... (https://linux.die.net/man/2/stat) You write the attributes via chmod(...) Similarly you can use chown(...) (https://linux.die.net/man/3/chown) As for the symbolic links: read: https://man7.org/linux/man-pages/man2/readlink.2.html creation: https://man7.org/linux/man-pages/man2/symlink.2.html

Full help: If you drop seppuku() from void Jidac::usage() it might work

fcorbelli commented 1 year ago

stat relies on a BIG structure (actually, two) with a LOT of data inside Storing something that big requires (to maintain backward compatibility) a lot of work Doable, in fact, but big (much bigger than -windate)

Restoring is hard too (a lot of strange things) Remember: zpaqfranz does not work "only" on Linux, but on Mac, BSD and Solaris and vSphere and NASes too. This file-system specific functions are a pain in the ass

Full help: without seppuku() you can get a LOT of help when inserting wrong parameters (even two or 3 times) I'll restore (much easier)

fcorbelli commented 1 year ago

Please check this one (and do not forget a review on sourceforge, please :) 56_2d.zip

ratay commented 1 year ago

I understand It is annoying that the original author outlined the work with attributes, but did not leave space in the structures for their future completion. Owner and group are really basic attributes without which a full backup of any unix-like system is not complete. It's a pity. But I believe that when you finish working on your current modifications, you will find some free time and space in "franz_block" to implement them. :) Also, symlink support should not be so complicated - I don't think there is any need to change the structures.

The full help seems to work now (useful for example to compare changes between versions) thanks

fcorbelli commented 1 year ago

It's a situation that I "feel" very little, and that's why I've never done it. I work as a system administrator: all files are ... mine. Mahoney has worked to achieve a very compact system, with very little wasted space That's why the different structures between windows and linux I am, on the other hand, much less interested in getting small files, also because I work with terabytes

ratay commented 1 year ago

I also work as a system administrator. And all files belong to ... someone :)) And as always, I found a workaround:

getfacl -R -P * > aclbackup.txt
setfacl --restore=aclbackup.txt
find /root -type l -print0 | tar --null --files-from - -cvf linkbackup.tar

BTW I tried the ESX version (ver. 5.5) and ended up like this:

zpaqfranz v55.15q-experimental-ESX archiver,  (Sep 19 2022)
franz:DEBUG very verbose (-debug)
21026: check_if_password of zp.zpaq
zp.zpaq:
0 versions, 0 files, 0 fragments, 0 blocks, 0 bytes (0.00 B)
Default sort
Updating zp.zpaq at offset 0 + 0
Adding 551 (551.00 B) in 1 files (0 dirs), 1 threads
20006: converting to localtime 2022-11-29 09:42:20
@ 2022-11-29 09:42:20
Real write
39481: g_cdatasize  0
39482: g_htsize     0
39431 writeJidacheader -1
23013: zpaqfranz error: St9bad_alloc

0.047 seconds (00:00:00)  (with errors)
33026: call xcommand on errorcode==2
Entering xcommand()
i_command   ||
i_parameter |Out of memory|
Exit because empty i_command
`
fcorbelli commented 1 year ago

Esx runs on a Centos 32 bit of about 15 years ago (!) does not allocate happlily RAM to anything else than VM On esx I use the r (robocopy/rsync) in a ghetto script and sum to make hashes A to store logs into a permanent datastore

I find 0 documentation (offical from vmware) for custom made exes If you have any please le me know

ratay commented 1 year ago

I also use ghetto - I thought I would supplement it with zpaq. Then why are you publishing the version for ESX when it doesn't work?

fcorbelli commented 1 year ago

Well it works for me, or better for my job

Everything from zpaq to zpaqfran are my very own 'pieces' of work I need this, I implement I do NOT need? I don't, unless someone else ask I do not need a zpaq compression into esxi: little free CPU, slow non cached filesystem,therefore no developing

Instead I need a sector based windows imager and a ransom-resilient nix

PS on Esxi 6.5 it compress (small) files,at least on mine servers

ratay commented 1 year ago

OK my misunderstanding (since nothing is published in the description) sorry