Thanks for your work on xen packages, I have some question about 4.5 packages: I saw a debian/rules that disable blktap2 but debian/patches/tools-blktap2-prefix.diff is still present and used in series, if blktap2 is now disable I think is nomore needed. I saw also many other patches about prefix still present but for xen 4.5 time ago I saw a big series of patches about honor tools prefix, probably some or all prefix patches are nomore needed. Can you update also xendomains init and default files in debian folder to latest xen upstream please? Some useful options added in upstream some years ago is still missed in debian packages :( Thanks for any reply and sorry for my bad english. -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 4297 bytes Desc: Firma crittografica S/MIME URL: <http://lists.alioth.debian.org/pipermail/pkg-xen-devel/attachments/20141210/bfe533ed/attachment.bin>
On Wed, 2014-12-10 at 12:47 +0100, Fabio Fantoni wrote:> Thanks for your work on xen packages, I have some question about 4.5 > packages: > I saw a debian/rules that disable blktap2 but > debian/patches/tools-blktap2-prefix.diff is still present and used in > series, if blktap2 is now disable I think is nomore needed.Yes, this can probably go.> I saw also many other patches about prefix still present but for xen 4.5 > time ago I saw a big series of patches about honor tools prefix, > probably some or all prefix patches are nomore needed.All which can be dropped have been dropped as part of the rebase AFAIK. Debian moves various things around more than the upstream work in this area supports (specifically we move lots of stuff from LIBDIR and *BINDIR to LIBEXEC).> Can you update also xendomains init and default files in debian folder > to latest xen upstream please? Some useful options added in upstream > some years ago is still missed in debian packages :(As I stated in response to your bug #760270 simply overwriting the current scripts with the current upstream ones is not the way to go, just asking for the same thing again isn't going to change that. Reducing the divergence, possibly even getting to the point where the Debian initscripts are just patches on top of the upstream ones, is something I would like to do, but not something I'm going to have time for any time soon. If you have specific issues with the Debian initscripts please report those as independent bugs. Ian.
Il 10/12/2014 12:56, Ian Campbell ha scritto:> On Wed, 2014-12-10 at 12:47 +0100, Fabio Fantoni wrote: >> Thanks for your work on xen packages, I have some question about 4.5 >> packages: >> I saw a debian/rules that disable blktap2 but >> debian/patches/tools-blktap2-prefix.diff is still present and used in >> series, if blktap2 is now disable I think is nomore needed. > Yes, this can probably go. > >> I saw also many other patches about prefix still present but for xen 4.5 >> time ago I saw a big series of patches about honor tools prefix, >> probably some or all prefix patches are nomore needed. > All which can be dropped have been dropped as part of the rebase AFAIK. > Debian moves various things around more than the upstream work in this > area supports (specifically we move lots of stuff from LIBDIR and > *BINDIR to LIBEXEC).Sorry, I saw this only after checking better now.>> Can you update also xendomains init and default files in debian folder >> to latest xen upstream please? Some useful options added in upstream >> some years ago is still missed in debian packages :( > As I stated in response to your bug #760270 simply overwriting the > current scripts with the current upstream ones is not the way to go, > just asking for the same thing again isn't going to change that. > > Reducing the divergence, possibly even getting to the point where the > Debian initscripts are just patches on top of the upstream ones, is > something I would like to do, but not something I'm going to have time > for any time soon. > > If you have specific issues with the Debian initscripts please report > those as independent bugs. > > Ian. >If I remember good there isn't particular bugs caused by the xendomains missed things. The more useful parameters when mainly windows domUs are used is XENDOMAINS_CREATE_USLEEP that I setted to 30 or 40 seconds on my systems to decrease the "overload" caused by windows startup (even if pv drivers are installed). The better performance of upstream qemu (in particular heavy disks performance increase with qdisk instead of blktap2) have minimized the problem but is still present starting 3-4 windows domUs or more. -------------- next part -------------- A non-text attachment was scrubbed... Name: smime.p7s Type: application/pkcs7-signature Size: 4297 bytes Desc: Firma crittografica S/MIME URL: <http://lists.alioth.debian.org/pipermail/pkg-xen-devel/attachments/20141210/5f48c922/attachment.bin>