qschulz has quit [Remote host closed the connection]
qschulz has joined #yocto
Vonter has quit [Ping timeout: 240 seconds]
Vonter has joined #yocto
davidinux has quit [Ping timeout: 252 seconds]
davidinux has joined #yocto
starblue has quit [Ping timeout: 255 seconds]
starblue has joined #yocto
xmn has joined #yocto
lthadeus has joined #yocto
Chocky has quit [Ping timeout: 264 seconds]
Chocky has joined #yocto
jclsn has quit [Ping timeout: 268 seconds]
jclsn has joined #yocto
lthadeus has quit [Ping timeout: 268 seconds]
khem has quit [Quit: Connection closed for inactivity]
xmn has quit [Quit: ZZZzzz…]
Chocky has quit [Quit: Leaving.]
Chocky has joined #yocto
alessioigor has joined #yocto
alessioigor has quit [Remote host closed the connection]
alessioigor has joined #yocto
jmd has joined #yocto
Omax has quit [Remote host closed the connection]
Chaser has joined #yocto
Omax has joined #yocto
wicki has joined #yocto
sakman has joined #yocto
Chaser has quit [Quit: Chaser]
<kanavin>
RP: note that both local and cdn tests fail in the same way, which means there's something yet unknown in qemuarm64 build which causes sstate mismatches with the test
<kanavin>
RP: the tricky part is to go back to the qemuarm64 build and see what objects it used, which is tricky because it gets erased on success :( so it needs to be caught in manually triggered sequence
<kanavin>
RP: and these seem the same objects we've had issues with in the previous iteration of the test which used printdiff
<kanavin>
RP: I suspect librsvg is the root, but somehow comparing signatures needs to be done
kpo has quit [Remote host closed the connection]
kpo has joined #yocto
<RP>
kanavin: I suspect hash equivalence does something we don't quite realise/understand :/
<kanavin>
RP: I've restarted qemuarm64 now, will pull out what it's used mid-flight (it has a long unaccelerated QA step in the middle)
<RP>
kanavin: also, why just package_write_rpm
<RP>
kanavin: not all the evidence is lost. For example, in that same build,https://autobuilder.yoctoproject.org/typhoon/#/builders/42/builds/8277/steps/12/logs/stdio has NOTE: Task /home/pokybuild/yocto-worker/qemuarm64/build/meta/recipes-gnome/librsvg/librsvg_2.56.3.bb:do_package_write_rpm unihash changed to 4865e77e122554d6b40cb3a0f4af5d817b4ba907c714dbbbff9a97955ff97f98
<RP>
kanavin: it is interesting that it changes the do_package unihash
<RP>
perhaps it changes the do_package unihash and never generates the rpm output?
<RP>
kanavin: we could probably verify that by computing hashes with the values forced to the one in that build log?
<RP>
kanavin: crazy idea - add a step to the end of the pkgman-non-rpm build which writes rpms?
<RP>
I am worried about why the hashes in that build are somehow differing to the other builds though
<kanavin>
RP: I'm running the manual test sequence now (qemuarm64 followed by yocto-mirror), let's see if I can pull out the differing signature values
<kanavin>
so it's that one we should look at (I clicked 'rebuild' on one in a-full)
<kanavin>
but it does seem to pull things from sstate too
<RP>
kanavin: yes :/
<RP>
kanavin: but that missing sstate is now there
<RP>
kanavin: so probably one of the other builds ended up generating it
lthadeus has joined #yocto
<kanavin>
RP: we really ought to start writing build ids into sstate siginfo files, that would contain clickable links to the autobuilder like ones above
<RP>
kanavin: not reproducible :/
<kanavin>
RP: why? the sstate object would not have them, just the siginfo file next to it
<RP>
kanavin: the siginfo files wouldn't be reproducible
<kanavin>
do they need to be?
<kanavin>
bitbake-diffsig and friends would specifically not consider the build id, and if it's otherwise identical, then it's reported as such
<RP>
kanavin: ideally
<RP>
kanavin: I'm going to try a tweak to helper, see if this does stop the CDN test failures
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
sakman has quit [Read error: Connection reset by peer]
<kanavin>
RP: going back to that qemuarm64 log in the original problematic build, I am seeing that rpms are being created, or do I misunderstand it?
<kanavin>
NOTE: Task /home/pokybuild/yocto-worker/qemuarm64/build/meta/recipes-graphics/xorg-font/font-util_1.4.1.bb:do_package_write_rpm unihash changed to 1a6894ac1a7eb8c48139c2f223eda3aa14131be218530b040a524b6376bc9824
sakman has joined #yocto
<RP>
kanavin: the pkgman-non-rpm build does not
<RP>
kanavin: imagine some new equivalence is found which means do_package changes for librsvg and ipk/deb are written but not rm
<RP>
rpm
sakman has quit [Ping timeout: 276 seconds]
<kanavin>
RP: right, yes
sakman has joined #yocto
<RP>
kanavin: it does seem strange :/
<landgraf>
Packaging patch I've sent few weeks ago caused some "base-files missing in sstate" failures on AB too. Still don't know how to investigate locally.
<kanavin>
landgraf, base-files missing is not a real failure, no need to worry
<landgraf>
kanavin: It was "No SPDX file found for package base-files, False sstate:base-files:qemux86_64-poky-linux" which failed the build.
<landgraf>
and SPDX part is even more "greek" for me than sstate :-(