1562804645 Q * _pa Read error: Connection reset by peer 1562811262 J * hygrometers ~sparklake@179.183.147.106 1562812134 M * Bertl_oO off to bed now ... have a good one everyone! 1562812137 N * Bertl_oO Bertl_zZ 1562813905 J * fstd_ ~fstd@xdsl-78-35-86-128.nc.de 1562814377 Q * fstd Ping timeout: 480 seconds 1562819377 Q * obeardly Ping timeout: 480 seconds 1562819932 J * obeardly ~obeardly@2603:3011:1661:0:9657:a5ff:feae:1552 1562825466 M * Guy- x1450: it's implemented using filesystem attributes, not xattrs, so it works with some filesystems but not others; extX works, jfs works; I think reiser maybe does too; xfs and zfs don't 1562828812 J * Ghislain ~Ghislain@211.ip-51-68-231.eu 1562830863 J * hijacker ~nikolay@149.235.255.3 1562836219 M * x1450 Ahh thanks, I'm guessing it would be BTRFS and ZFS only then? 1562836353 M * x1450 the issue i'm trying to solve is that apt-get occasionally hits errors updating because it can't overwrite files that have been hashified however excluding them would also result in less effective space savings 1562842558 M * CcxWrk xfs now has it's own COW thing though, so you could just use that 1562842673 M * CcxWrk (xfs reflinks are stable as of 4.16) 1562844016 M * x1450 on ext4 :( 1562844539 J * _pa ~pav@141-136-157-48.dsl.iskon.hr 1562845580 Q * Aiken Remote host closed the connection 1562846623 M * Guy- x1450: no, it works with ext4; it *doesn't* work with zfs 1562846671 M * Guy- x1450: however, I also had a box where I hit a similar problem with earlier kernels or earlier util-zfs versions; the iunlink attribute was not set during hashification, only the immutable attribute 1562847774 M * x1450 Guy-, I'm using the 4.9 kernel at the moment, what was your resolution for it? 1562848985 M * Guy- x1450: this was ages ago, with perhaps a 4.1.x kernel 1562849003 M * Guy- no, must have been even earlier 1562849058 M * Guy- x1450: however, I had a daily cronjob to workaround the problem 1562849167 M * Guy- x1450: this was the script: https://pastebin.com/mMNr1X5f 1562849205 M * Guy- don't use it without understanding how it works/what it does 1562849933 N * Bertl_zZ Bertl 1562849938 M * Bertl morning folks! 1562850727 M * x1450 so based on that, you recursively set the ilink after the hashify process? then run a cleanup 1562850744 M * x1450 if that's all that is required, should be a simple modification on my end 1562852906 M * Bertl any filesystem which has block level deduplication (like BTRFS) has no need for COW link breaking 1562852936 M * Bertl and yes, you basically create a hard link, set the proper flags and you're done 1562852964 M * Bertl when the kernel encounters the flags and sees a hard link, it copies the content and removes the hard link 1562852995 M * Bertl off for now ... bbl 1562852999 N * Bertl Bertl_oO 1562853512 M * x1450 thanks Guy- and Bertl_oO 1562854560 M * Guy- x1450: yes 1562854591 M * Guy- Bertl_oO: but, even with block level dedup, won't libraries with different inode numbers but deduplicated blocks be mapped twice? 1562855551 Q * hygrometers Ping timeout: 480 seconds 1562856413 Q * Ghislain Quit: Leaving. 1562857715 J * Ghislain ~Ghislain@211.ip-51-68-231.eu 1562861834 Q * hijacker 1562862914 M * Bertl_oO Guy-: correct, but that's where KSM can help 1562866006 J * hijacker ~nikolay@2001:981:4551:1:9504:b68f:a87f:a673 1562868082 Q * hijacker Remote host closed the connection 1562875923 J * Aiken ~Aiken@b951.h.jbmb.net 1562876066 Q * _pa Quit: Leaving 1562876089 J * _pa ~pav@141-136-157-48.dsl.iskon.hr