Re: [blfs-dev] About LVM test hanging
Em 26-10-2015 18:11, Pierre Labastie escreveu: > As noted in ticket #6931, one test (test/shell/lvcreate-large-raid.sh) hangs > forever and may even trash the system, which then needs a hard reboot. > > I've found the explanation, but really upstream is too "fedora-centric" (see > below why). In file test/lib/aux, there is this function (look at the 4 line > long comment): > > have_raid() { > test "$RAID" = shared -o "$RAID" = internal || { > echo "Raid is not built-in." >&2 > return 1; > } > target_at_least dm-raid "$@" > > # some kernels have broken mdraid bitmaps, don't use them! > # may oops kernel, we know for sure all FC24 are currently broken > # in general any 4.1, 4.2 is likely useless unless patched > # hopefully 4.3 will be patched > case "$(uname -r)" in > 4.[123].*fc24*) return 1 ;; > esac > --- > The comment shows that mdraid is broken in kernels version 4.1 and 4.2, but > then, only fc24 kernels lead to non zero exit! Note that according to the > comment, any test using raid could be hanging too. I've only observed that for > one test. > > Anyway, I tried linux-4.0.9 and linux-4.3-rc7, and the test passes on both. > Looks like 4.3 has been patched. > I'll change the book to warn against versions 4.1 and 4.2 of the kernel (in > both LVM and mdadm). Yes, good change for all users end developers. And I left one comment in the ticket. Thanks again for your deep investigation about these so problematic checks. -- []s, Fernando, aka SÃsifo -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-dev] About LVM test hanging
On 27/10/2015 15:45, Fernando de Oliveira wrote: > Em 26-10-2015 18:11, Pierre Labastie escreveu: >> As noted in ticket #6931, one test (test/shell/lvcreate-large-raid.sh) hangs >> forever and may even trash the system, which then needs a hard reboot. >> >> I've found the explanation, but really upstream is too "fedora-centric" (see >> below why). In file test/lib/aux, there is this function (look at the 4 line >> long comment): >> >> have_raid() { >> test "$RAID" = shared -o "$RAID" = internal || { >> echo "Raid is not built-in." >&2 >> return 1; >> } >> target_at_least dm-raid "$@" >> >> # some kernels have broken mdraid bitmaps, don't use them! >> # may oops kernel, we know for sure all FC24 are currently broken >> # in general any 4.1, 4.2 is likely useless unless patched >> # hopefully 4.3 will be patched >> case "$(uname -r)" in >> 4.[123].*fc24*) return 1 ;; >> esac >> --- >> The comment shows that mdraid is broken in kernels version 4.1 and 4.2, but >> then, only fc24 kernels lead to non zero exit! Note that according to the >> comment, any test using raid could be hanging too. I've only observed that >> for >> one test. >> >> Anyway, I tried linux-4.0.9 and linux-4.3-rc7, and the test passes on both. >> Looks like 4.3 has been patched. >> I'll change the book to warn against versions 4.1 and 4.2 of the kernel (in >> both LVM and mdadm). > > Yes, good change for all users end developers. > > And I left one comment in the ticket. > > Thanks again for your deep investigation about these so problematic checks. > > I've a question. I'd like to put the warning in an admonition, but I do not know whether I should use "note", "caution", "warning", or something else. I'd go for caution but maybe one of you knows better. BTW, I tried linux-4.2.5 today, and it is patched too. Pierre -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-dev] About LVM test hanging
Pierre Labastie wrote: I've a question. I'd like to put the warning in an admonition, but I do not know whether I should use "note", "caution", "warning", or something else. I'd go for caution but maybe one of you knows better. A warning or caution would be if you could damage your system. A note would be appropriate here since the hanging test can just be killed. BTW, I tried linux-4.2.5 today, and it is patched too. Good. -- Bruce -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-dev] About LVM test hanging
On 27/10/2015 22:03, Bruce Dubbs wrote: > Pierre Labastie wrote: > >> I've a question. I'd like to put the warning in an admonition, but I do not >> know whether I should use "note", "caution", "warning", or something else. >> I'd >> go for caution but maybe one of you knows better. > > A warning or caution would be if you could damage your system. A note would > be appropriate here since the hanging test can just be killed. Hmm, not sure it is really "damage", but I had to hard reboot several times... Anyway, will do a note, thanks. Pierre -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
Re: [blfs-dev] About LVM test hanging
On 27/10/2015 22:09, Pierre Labastie wrote: > On 27/10/2015 22:03, Bruce Dubbs wrote: >> Pierre Labastie wrote: >> >>> I've a question. I'd like to put the warning in an admonition, but I do not >>> know whether I should use "note", "caution", "warning", or something else. >>> I'd >>> go for caution but maybe one of you knows better. >> >> A warning or caution would be if you could damage your system. A note would >> be appropriate here since the hanging test can just be killed. > > Hmm, not sure it is really "damage", but I had to hard reboot several times... > > Anyway, will do a note, thanks. > > Pierre > Done at r16572. Not finished with that yet. I added a comment to the ticket too, because I am not sure what Fernando has in mind. Pierre -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page
[blfs-dev] About LVM test hanging
As noted in ticket #6931, one test (test/shell/lvcreate-large-raid.sh) hangs forever and may even trash the system, which then needs a hard reboot. I've found the explanation, but really upstream is too "fedora-centric" (see below why). In file test/lib/aux, there is this function (look at the 4 line long comment): have_raid() { test "$RAID" = shared -o "$RAID" = internal || { echo "Raid is not built-in." >&2 return 1; } target_at_least dm-raid "$@" # some kernels have broken mdraid bitmaps, don't use them! # may oops kernel, we know for sure all FC24 are currently broken # in general any 4.1, 4.2 is likely useless unless patched # hopefully 4.3 will be patched case "$(uname -r)" in 4.[123].*fc24*) return 1 ;; esac --- The comment shows that mdraid is broken in kernels version 4.1 and 4.2, but then, only fc24 kernels lead to non zero exit! Note that according to the comment, any test using raid could be hanging too. I've only observed that for one test. Anyway, I tried linux-4.0.9 and linux-4.3-rc7, and the test passes on both. Looks like 4.3 has been patched. I'll change the book to warn against versions 4.1 and 4.2 of the kernel (in both LVM and mdadm). Pierre -- http://lists.linuxfromscratch.org/listinfo/blfs-dev FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page