1529368451 J * Ghislain1 ~ghislain@adsl1.aqueos.com
1529368451 Q * Ghislain Read error: Connection reset by peer
1529377665 M * Bertl_oO  off to bed now ... have a good one everyone!
1529377667 N * Bertl_oO Bertl_zZ
1529395282 J * nikolay ~nikolay@149.235.255.3
1529400250 J * Gremble ~Gremble@cpc1-aztw34-2-0-cust397.18-1.cable.virginm.net
1529401529 N * Bertl_zZ Bertl
1529401531 M * Bertl  morning folks!
1529401589 M * gnarface  morning Bertl!
1529403615 M * Ghislain1 heelo you, i must go eat but i will not go without doing the 4.9 update mantra first ;p
1529404093 M * Bertl great! please go ahead!
1529404135 M * Bertl btw, it looks like I might finally find the time today ... but as usual, no guarantees
1529404553 M * Ghislain1 lol
1529404560 M * Ghislain1 that would be great
1529404573 M * Ghislain1 i seen kernel guys continue their daily deliveries :)
1529404594 M * Ghislain1 4.9.24598004 is soon to be released !:D
1529404822 M * Bertl yeah
1529404854 M * Bertl btw, I'm surprised how good btrfs is nowadays
1529404897 M * Bertl it has a lot of Linux-VServer friendly features like CoW and deduplication as well as lightweight snapshots
1529404959 M * Bertl seems to provide some advanced quota system as well
1529408162 M * Ghislain1 what would it need to have the quota syb system to work on vserver ?
1529408169 M * Ghislain1 s/syb/sub/
1529408206 M * Ghislain1 as i use shared space my use need to be able to put quotas, on user if i can, on directory if not
1529408316 M * Bertl the quota stuff has separate quota groups and you can have separate subvolumes too
1529408333 M * Bertl (and still use CoW and reflink)
1529408394 M * Bertl haven't tried the quota stuff yet, might need some 'blocking' capabilities or so, but on paper it looks solid and useable
1529408893 Q * Aiken Remote host closed the connection
1529410608 Q * Gremble Quit: Leaving
1529412746 M * Ghislain1 hum ok next kernel compile i will include it and see if i can test the beast
1529412764 M * Ghislain1 as long as raid 5/6 is not used i think it is quite solid now
1529412789 M * Ghislain1 main contrib is facebook and they use only raid1/10, the other levels are quite not tested
1529413174 M * Bertl yeah, but I don't see a problem with staying away from btrfs based raid 
1529413198 M * Bertl you can easily run it on soft raid (lvm/mdadm)
1529413210 M * Guy- yes, but then you lose all benefits of its own "raid"
1529413230 M * Guy- such as self-healing, or (I assume) variable stripe size (zfs has that)
1529413233 M * Bertl yeah, but I'm not sure that there are many benefits :)
1529413247 M * Guy- I only know about zfs; in that case, there are many benefits
1529413268 M * Guy- for example, resilvering only syncs sectors that are actually in use
1529413285 M * Guy- regular raid has no idea which sectors are in use, so it has to resync everything
1529413295 M * Bertl those are corner cases at best
1529413329 M * Guy- self-healing (knowing which of two different but supposedly identical data elements is the correct one) is hardly a corner case
1529413367 M * Guy- variable length raid blocks also provide large performance benefits, especially in the raid5/6 case (because it avoids read-modify-write)
1529413373 M * Guy- not to mention avoiding the "write hole"
1529413562 M * Ghislain1 on my side i only user raid1 and 10 so this is not a issue for me. Of course larger disk array (i have 2 or 4 here) could waste space not having the likely raidz2 of zfs
1529413580 M * Ghislain1 but zfs is not included in the kernel 
1529413588 M * Guy- nor will it ever be
1529413603 M * Guy- but these days it's easy to install, so it's not a big deal
1529413669 M * Ghislain1 so we have btrfs or the futur bcachefs
1529413710 M * Ghislain1 hum i really dont trust distro that run early 1990 zfs code to ship good version especialy when i dont run the distrib's kernel but the vserver ones
1529413718 M * Ghislain1 allready to many things to manage
1529415208 M * Guy- what early 1990 zfs code? :)
1529415222 M * Guy- Debian has fairly recent zfs (it's only a few weeks old)
1529415261 M * Guy- it doesn't depend on a specific kernel -- I run it on all my physical servers, all with vserver kernels
1529415912 M * Ghislain1 debian has most paquets dating 5 to 10 years back, it was a general statement ;p
1529415944 M * Ghislain1 having the utils not in sync with the kernel do not ring like a good idea to me but this is my spider sense not a reality check
1529417536 M * Guy- there is no kernel dependency in the utils
1529417568 M * Guy- the zfs utilities may depend on minimum versions of the zfs kernel module, but not on the version of the kernel
1529417590 M * Guy- and I promise you the zfs-dkms package in Debian is fairly recent :)
1529417817 M * Ghislain1 you nose moved i know you lie ! :)
1529417873 M * Guy- I always lie
1529417873 M * Ghislain1 i tried to see how to compile the kernel with zfs but this seemed quite confusing and daily stuff caugth so i did not pursue
1529417876 M * Ghislain1 lo
1529417880 M * Guy- in fact, I'm lying to you right now!
1529417894 M * Guy- you don't need to compile the kernel with zfs
1529417911 M * Guy- you just install the zfs-dkms and spl-dkms package (and zfsutils-linux, for the utilities)
1529417980 M * Guy- (compiling the kernel with zfs built in is possible, but in general the dkms approach is better)
1529418087 M * Bertl what I don't like about ZFS is the complexity and the fact that it seems to eat memory for breakfast, lunch and dinner :)
1529418360 M * Ghislain1 very powerfull but complex yes. It has some wicked thing with compressed ark for exemple
1529418606 M * Guy- it certainly benefits from a large cache (that is, lots of memory); this is due to it being a COW filesystem, which causes files to be fragmented
1529418620 M * Guy- but I've used it on a netbook with just 4GB of RAM :)
1529418668 M * Guy- it's certainly more complex than, say, ext2, but it has many benefits you pay for with the added complexity
1529418756 M * Guy- feature-wise, btrfs seems to be pretty much a knock-off of zfs :) (and there's nothing wrong with that; it's also not surprising considering that Valerie Aurora used to be involved with zfs before she went on to write much of btrfs)
1529418759 M * Bertl hah! didn't know that there was a zfs-fuse
1529418767 M * Guy- stay away from that
1529418775 M * Guy- it's old, unmaintained, slow and unstable
1529418804 M * Bertl yeah, what else do you expect from a fuse filesystem :)
1529418842 M * Guy- it doesn't have to be old and unmaintained, surely :)
1529418868 M * Bertl the fuse interface kind of encourages this though
1529419018 M * Guy- I'll take your word for that
1529421316 M * Ghislain1 the btrfs difference is that it simplified  the inner working by using a central c tree  system where zfs has many diffrent ubsystems, but i got this by an old reading on the btrfs birth sdo dont  quote me on that
1529421330 M * Ghislain1 b tree not c tree lol
1529424764 Q * nikolay Quit: Leaving
1529440214 J * Aiken ~Aiken@2001:44b8:2168:1000:b26e:bfff:fe2a:b951
1529443560 M * Bertl  off to bed now ... have a good one everyone!
1529443561 N * Bertl Bertl_zZ
1529444153 J * Gremble ~Gremble@cpc1-aztw34-2-0-cust397.18-1.cable.virginm.net
1529447275 Q * Gremble Quit: Leaving
1529448427 J * fstd_ ~fstd@xdsl-85-197-61-39.netcologne.de
1529448875 Q * fstd Ping timeout: 480 seconds
1529448875 N * fstd_ fstd