dgilmore changed the topic of #fedora-riscv to: Fedora on RISC-V https://fedoraproject.org/wiki/Architectures/RISC-V || Logs: https://libera.irclog.whitequark.org/fedora-riscv || Alt Arch discussions are welcome in #fedora-alt-arches
iooi has quit [Quit: iooi]
iooi has joined #fedora-riscv
tg has quit [Quit: tg]
tg has joined #fedora-riscv
warren has left #fedora-riscv [Leaving]
jcajka has joined #fedora-riscv
<Entei[m]> Koji web is unable to establish connection to koji hub.
<Entei[m]> I am able to establish connection from client to hub
<kwizart> did you tried to open firewall port or enable the right selinux boolean ?
<rwmjones> davidlt[m]: that gcc build is still going!
<rwmjones> I kind of wonder if it's short of RAM
<rwmjones> it's not swapping
<davidlt[m]> How many days is it going now?
<davidlt[m]> Entei: there might be more rules, for example SELinux.
<davidlt[m]> Check all the logs incl. SElinux reports.
<Entei[m]> <davidlt[m]> "Entei: there might be more rules..." <- Probably. Will look into it
<Entei[m]> Meanwhile I'd like to try building packages through the CLI. Looked into the comps, the Buildsystem group list couple of essential packages. But it doesn't feel like a complete list. It lists `gcc` and `make`, but not `binutils` or `glibc`?
<Entei[m]> That container from mock will need certainly need those to build packages
<davidlt[m]> That's fine, dependencies will pull those.
<davidlt[m]> Koji comps are minimal. Dump the list from upstream Koji or ours and define those.
<davidlt[m]> Those just need to be copy & pasted.
<davidlt[m]> Koji will generate mock config and will tell it to create chroot using those groups.
<davidlt[m]> Different groups for different tasks.
zsun has joined #fedora-riscv
<Entei[m]> <davidlt[m]> "That's fine, dependencies will..." <- But I need to populate my repository with those dependency packages. Where can I find the whole list?
<davidlt[m]> You need to seed your Koji with initial RPMs.
<davidlt[m]> That would be importing our repositories as-is and whatever you have now on top.
<davidlt[m]> Then it's all about mass rebuilding in multiple stages.
<conchuod> Seems like quite the miserable process...
<davidlt[m]> It is, but getting rid of C is somewhat easy.
<davidlt[m]> It's just throwing a bunch of compute at it mainly.
<conchuod> Most of this sounds less like chucking compute & more like doing fumbling through some underdocumented setup
<davidlt[m]> Oh yeah.
<davidlt[m]> That's why there are infra team normally 😀
<davidlt[m]> s/are/is/
<davidlt[m]> Multiple folks working on keeping the machinery running.
<Entei[m]> davidlt: are you running qemu from rawhide repo? I noticed some hang ups during boot with my built binary, finally installed qemu-8.0 from rawhide repo, but still unable to boot with more than 8 cores....
<Entei[m]> I know you are running on bare-metal for builders right now, but if you have any emulators running with more than 8 cores do let me know
<conchuod> Where do you hang during boot?
<Entei[m]> conchuod: starting gdm. Running with -nographic btw
<Entei[m]> <Entei[m]> "davidlt: are you running qemu..." <- Ignore this. Could have been a cache problem or something. Now I am magically able to boot as desired. Fedora's built qemu binary makes the image feel quite a bit more snappy compared to what I built
sxa[m] has joined #fedora-riscv
<Entei[m]> Btw, I see you did a gcc build on 23rd June and finished on 2nd Jul. Was this on Unmatched boards?
<davidlt[m]> Yes
<davidlt[m]> Unmatched
<Entei[m]> <davidlt[m]> "Unmatched" <- Damn it's actually less time to build on 4 real cores then 20 emulated cores. That's impressive. I think I should request some LicheePi4a boards or similar for builders.
<Entei[m]> How do the CPUs on those non-SiFive boards compare to Unmatched boards? Any ideas on that?
zsun has quit [Quit: Leaving.]
zsun has joined #fedora-riscv
<sxa[m]> The cores on Lichee4A/VF2 are definitely quite a bit quicker than UNmatched
zsun has quit [Quit: Leaving.]
jcajka has quit [Quit: Leaving]
tg has quit [Quit: tg]