Many changes. Reproducible: Always
Created attachment 265845 [details] ebuild for new releas 0.7.0 working ebuild w/ small warnings regarding use of "BuildDir()" in SConscript files. init-script files/zfs-fuse is the same as in the current 0.6.9-r1 in portage.
Created attachment 265849 [details] updated ebuild for 0.7.0 removed warnings by patching the SConscript-files.
Created attachment 266009 [details] checking out the already patched maint-branch out of their git-repo It builds OK here without those scons-warnings, pls someone test it ...
Created attachment 266023 [details] zfs-fuse-0.7.0 maint-branch, correction regarding initd-script
As i can see this is live ebuild, not ebuild for version 0.7.0 . And still doesn't respect CFLAGS/LDFLAGS/CC. Patch from bug #332395 , with small changes, could solve this problem. (btw, i'll try refresh those patch).
(In reply to comment #5) > As i can see this is live ebuild, not ebuild for version 0.7.0 . And still > doesn't respect CFLAGS/LDFLAGS/CC. Patch from bug #332395 , with small changes, > could solve this problem. (btw, i'll try refresh those patch). Marcin, pls check #362431 as well, I tried to add your patch there.
*** Bug 362431 has been marked as a duplicate of this bug. ***
Just bumping: Who waits for whom to do what to get zfs-fuse-0.7.0 into portage?
been running like stable for me for months off my x86 overlay. not that this answers the bump question. :)
(In reply to comment #9) > been running like stable for me for months off my x86 overlay. not that this > answers the bump question. :) stable here as well for months, just as feedback to the devs. bump-q still open, yes ;-)
*** Bug 398217 has been marked as a duplicate of this bug. ***
What is keeping this out of the main tree?
Weird. How come this is still not in the tree, after nearly a year? @"Christian Parport": Is there something we can do to help?
(In reply to comment #13) > Weird. How come this is still not in the tree, after nearly a year? > > @"Christian Parport": Is there something we can do to help? Samuli Suominen masked sys-fs/zfs-fuse-0.6.9 for removal around the time of my previous comment citing several bug reports, including this one. /usr/portage/profiles/package.mask: # Samuli Suominen <ssuominen@gentoo.org> (09 Jan 2012) # Fails to compile against recent glibc and multiple other issues. # No maintainer to take care of the package. Removal in 30 days. # Bugs 325331, 284369, 380749, 388741, 381539, 358697, and 332395. At this point, I have mostly switched to using the 9999 ebuild for the ZFS kernel modules from the pinkbyte overlay, but perhaps this could be made a tracker bug for the inclusion of sys-fs/zfs-fuse-0.7.0. That would enable users of ZFS-Fuse to coordinate efforts to resolve the issues here.
(In reply to comment #14) > At this point, I have mostly switched to using the 9999 ebuild for the ZFS > kernel modules from the pinkbyte overlay[.] Just tried that, and > configure: error: > *** Kernel built with CONFIG_PREEMPT which is not supported. > ** You must rebuild your kernel without this option. So it’s useless on any modern system. :/ (The “sping” overlay has a zfs-9999 too, but it also doesn’t compile, and for harder to fix reasons.) Seems there is no ZFS on Linux. :(
(In reply to comment #15) > (In reply to comment #14) > > At this point, I have mostly switched to using the 9999 ebuild for the ZFS > > kernel modules from the pinkbyte overlay[.] > > Just tried that, and > > configure: error: > > *** Kernel built with CONFIG_PREEMPT which is not supported. > > ** You must rebuild your kernel without this option. > > So it’s useless on any modern system. :/ > (The “sping” overlay has a zfs-9999 too, but it also doesn’t compile, and for > harder to fix reasons.) > > Seems there is no ZFS on Linux. :( That is issue 83 at the ZFS On Linux bug tracker: https://github.com/zfsonlinux/zfs/issues/83 As a workaround, you can set CONFIG_PREEMPT_VOLUNTARY. That setting was introduced after CONFIG_PREEMPT wiith the claim that it made things better for people doing low-latency audio work. I have been using it for a while with no discernible increase in desktop performance, so I suspect that you could use it until the issue in the ZFS kernel module is fixed.
(In reply to comment #16) > As a workaround, you can set CONFIG_PREEMPT_VOLUNTARY. That setting was > introduced after CONFIG_PREEMPT wiith the claim that it made things better for > people doing low-latency audio work. I have been using it for a while with no > discernible increase in desktop performance, so I suspect that you could use it > until the issue in the ZFS kernel module is fixed. Very good. :) I do audio work, but only on Windows (because of the lack of the software on Linux :/), so this is welcome. :)
Just as a clarification, I meant to write "no discernible decrease in desktop performance", not "no discernible increase in desktop performance". I realized this after Navid's response. Also, while I am using CONFIG_PREEMPT_VOLUNTARY on my desktop, I have not yet switched to using ZFS as my root filesystem. I have been working toward getting a server with advanced format drive to boot off a raidz2 array with ashift=12.. Zachary Bedell did some work on getting Gentoo to boot off ZFS last year, but his overlay is mostly broken with the exception of his genkernel-9999 ebuild. Currently, there are numerous issues that occur when trying to do a new stage3 install on ZFS. After I have my server booting off ZFS, I will try writing some documentation for people interested in booting off ZFS.
*** Bug 284369 has been marked as a duplicate of this bug. ***
(In reply to comment #18) BTW: I noticed there is a full CONFIG_PREEMPT. Why not select that on a desktop system?
(In reply to comment #20) > (In reply to comment #18) > BTW: I noticed there is a full CONFIG_PREEMPT. Why not select that on a desktop > system? CONFIG_PREEMPT and CONFIG_PREEMPT_VOLUNTARY take different approaches toward implementing kernel preemption. Right now, the ZFSOnLinux code is incompatible with the approach taken by CONFIG_PREEMPT.
This bug should be closed with OBSOLETE as zfs is in the tree as sys-fs/zfs and sys-fs/zfs-kmod
