02:19:35 It's too bad you hadn't just mounted swap over it instead. That's more often less of a problem. :-) 02:20:06 Oops! That was meant elsewhere. Move along, move along, nothing to see here. 03:03:18 * ek stares with a raised eyebrow at rwp 03:13:46 ek, A friend on Linux ZFS tried to add a 3rd disk into a mirror with the intention of sync'ing and removing the 2nd disk. But accidentally striped it into the array instead. Into the boot bpool and then said grub-zfs made it impossible to remove. No checkpoint. Problem with the recovery. Eventually the entire problem got worse. 03:18:12 rwp: That sounds about right. 03:18:27 Gotta be careful with them ZFS's. 03:18:57 I also am not a fan of how Linux does the ZFS for boot pool and root pool. I get it, but why? 03:20:02 I can't figure that out either. 03:20:41 I can see wanting to not have the boot on zfs and so separating it from the root rpoot. But then they put it on a boot bpool. Which is basically exactly the same thing. I can't figure it out either. 03:22:50 Exactly my thoughts. It's already there. Why separate /boot and /(root)? Maybe it's the way they update the bootloader? Dunno. 03:24:38 Given what was said about grub-zfs limitations I expect the reason for the bpool is also related in there somehow. 03:28:15 rwp: Very likely. 03:58:30 On the struggle bus 03:58:59 I'm still on the same kernel and userland from the install. I've done buildworld and buildkernel -- but not installed either 03:59:01 Hopefully, the short struggle bus. 03:59:13 ... wait. I dunno if that would be better or worse? 03:59:19 yet now my system will now hang on boot unless I choose "safe mode" 03:59:42 makes no sense whatsoever 03:59:55 sponix2ipfw: Likely confused with the kernel. You didn't installkernel before the reboot? 04:00:20 exactly what does "safe mode" boot do. It still loads my network and amdgpu on boot just fine 04:00:54 ek : nope, same kernel, the one I built but didn't install yet is GENERIC-NODEBUG 04:02:16 just get a black/blank screen when attempting to load X also... Even as root, just a blinking underscore in the top left 04:02:18 sponix2ipfw: And what kernel is booting when you're in safe mode? I'd imagine it's the original kernel. 04:02:32 FreeBSD buildbox 15.0-PRERELEASE FreeBSD 15.0-PRERELEASE #0 main-n279771-027be99b1f33: Thu Aug 21 20:15:59 UTC 2025 root⊙rnfo:/usr/obj/usr/src/amd64.amd64/sys/GENERIC amd64 1500063 1500063 04:02:42 sponix2ipfw: In single-user mode? That wouldn't surprise me. 04:02:43 same kernel 04:02:59 nope, I have multi-user in safe mode also 04:04:14 I'm assuming (hoping?) you're on ZFS and took a snapshot beforehand. If so, I'd do the installkernel, reboot, then install world, then the rest of the standard steps. 04:04:24 If that doesn't work, rollback and try again. 04:06:13 safe mode is supposed to turn off devmatch and disable a few other things 04:08:33 what is devmatch? 04:10:33 going to reboot again 04:11:50 Oh. He's IRC'ing from that same box, I guess. 04:12:18 Err... They? I dunno if they're a he. Be patient with me. I'm from the 1900's. 04:13:51 lol 04:14:23 booted all the way up in normal mode after installing the kernel -- the exact same generic kernel, except without debugging 04:14:29 FreeBSD buildbox 15.0-PRERELEASE FreeBSD 15.0-PRERELEASE #0: Sun Aug 24 21:26:04 CDT 2025 root@buildbox:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG amd64 1500063 1500063 04:14:42 no safe mode needed now 04:16:34 maybe this makes sense to everyone else. But I'm just lost as to why it does or doesn't work at this point 04:18:06 a beer or two might help? 04:18:14 always works for me 04:18:22 (beer) 04:18:31 beeeeeeer 04:28:22 lol... no it is back to only booting in safe mode again 04:28:28 FreeBSD buildbox 15.0-PRERELEASE FreeBSD 15.0-PRERELEASE #0: Sun Aug 24 21:26:04 CDT 2025 root@buildbox:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG amd64 1500063 1500063 04:28:50 define 'hang on boot' 04:29:03 same kernel, userland/installworld is from the same source that was working during the install 04:29:19 kevans it hits the ACPI line and freezes right there 04:29:27 if not booting in safe mode 04:29:53 I had this prior when having any USB legacy support going, turning all that off "fixed it" for a while 04:30:08 did you `boot -v`? 04:30:30 rtprio probably not, I'm not sure how to even do that 04:31:02 drop to the loader prompt and `boot -v`, or set boot_verbose="YES" in loader.con 04:31:06 f 04:31:22 All Vodka sucks no matter how they market it. This stuff says extra smooth. It still burns 04:31:56 okay, boot_verbose='YES" in boot/loader.conf it is 04:33:45 guess I will take a picture on my phone of where it hangs on boot, and then post that here 04:34:09 Normally on my decade old platform stuff "just works". This FreeBSD Current stuff has been a headache lately 04:34:46 brb, rebooting again 04:37:13 Boot hang https://usercontent.irccloud-cdn.com/file/iDx9LfIB/1756183016.JPG 04:40:50 okay, I'm getting the impression it just hangs on boot with a "reboot". And only boots up clean after a hard power down 04:40:55 FreeBSD buildbox 15.0-PRERELEASE FreeBSD 15.0-PRERELEASE #0: Sun Aug 24 21:26:04 CDT 2025 root@buildbox:/usr/obj/usr/src/amd64.amd64/sys/GENERIC-NODEBUG amd64 1500063 1500063 04:41:03 and NO I don't have fastboot enabled 04:41:24 it just booted up without safe mode, after a hard power down 04:41:56 I'm going to try "poweroff" instead of "reboot" and see if that proves true 04:45:25 Yeah, that is the problem. If I poweroff instead of rebooting it works fine 04:46:12 Now I just need to figure out why x11 doesn’t work 04:47:50 what's the context around where it hangs on a non-safe-mode functional boot? 05:00:03 okay, just need to make sure I don't do any soft reboots and only hard poweroffs, and I should be good there 05:00:16 I now can load KDE Plasma, but have an invisible mouse cursor 05:00:41 AMD RX 5700 GPU 05:19:39 Okay, I'm finally up and in X 05:20:01 this FreeBSD Current Prerelease really is FUN 05:20:12 thanks for the help everyone 06:25:13 vishwin welcome back 10:55:22 A little inquiry: are there any netcup users around here? Cause I've been looking into switching my tiny personal VPS at Hetzner over to one of their root servers. Price is good but I'd like to hear about your experience before I make the move. Thanks 15:49:20 in freebsd, when i unlock/decrypt a disk in the boot loader, it loads the system without asking for the password again. can this be used to reboot without reentering the password? 15:56:45 no. because reboot means memory wipe. So unless you want to store password into the known location and since it effectively needs to be plaintext, you can just as well use paper sticker on desk;) 16:20:50 gonna try freebsd on my htpc because debian hdmi audio is buggy 16:21:22 anyone want to be whether it will fix the problem? 16:26:20 Why not try to fix it first? There are some very helpful people over a #debian. Jumping to another OS because of a small issue is not a solution 16:27:00 the problem is that the bug is not easily reproducible 16:27:14 it shows up after a few hours or a few days of use, then disappears on reboot 16:28:11 In any way, #debian should be your first stop ;) 16:31:51 I have low confidence in fixing a bug that I can't reliably trigger 16:31:59 it will be useful to know if it's still a problem on freebsd anyway 16:33:21 Only you can tell 18:00:19 is there a way to make poudriere built with multi-thread on electron/chromium ? These are taking forever to build on my box, because they are trying to do so with only one thread on my 20-core CPU 18:02:12 never mind the LLM answered me 18:02:17 eh 18:02:22 will give it a shot 18:28:30 That LLM guy is smart. . . until he isn't. 18:30:21 CrtxReavr: That's pretty spot-on from what I've seen. 18:32:00 LLMs either impress me or have me shaking my head. There's not really any middle-ground. 19:11:23 so I'm seeing this in the handbook: 19:11:25 dd if=FreeBSD-13.1-RELEASE-amd64-memstick.img of=/dev/da0 bs=1M conv=sync 19:12:06 is da0 the UDB device itself, or the first partition? 19:12:15 s/D/S/ 19:12:34 like is da0 akin to sdc1 on linux, or sdc? 19:14:38 da0 is the disk 19:14:52 da0s1 or da0p1 is the partition, depending on partitioning 19:15:24 cool thanks 19:28:53 bleb, file -s /dev/da0 19:29:09 (That should answer your question.) 19:36:30 ty 19:37:25 It also works on partitions and should identify things like MBR or Gpart layouts, OS boot sectors, etc. 20:22:19 I installed freebsd on a partition but the grub menu from debian isn't showing it 20:22:40 I tried running update-grub2 on debian which runs os-prober, but it's not finding freebsd 20:22:47 any tips for getting freebsd to show up in the grub menu? 20:23:09 bleb desktop computer? 20:23:53 yes 20:32:47 bleb if you can scrape up $25, I recommend hooking up an SSD specifically all for FreeBSD. Then using the UEFI boot menu to select it at POST/Boot time 20:32:57 It will make your life a LOT easier 20:36:37 I am a fan of separate boot devices too 20:36:43 but how hard can it really be 20:36:58 different linux distros and windows 10 work fine with grub 20:37:12 on the same disk I mean 20:43:51 FreeBSD have boot manager that works with Gpart? In the MBR days, it has some snazzy features. 20:45:10 i don't consider "f1 freebsd f2 windows " very snazzy 20:46:21 Actually, you could select any primary partition on any disk wit it. 20:46:54 And it would default to the last partition booted. . .which was very handy in labs for remote reboots. 20:47:20 And I know. . .I used to make a metric shit-ton of of multi-boot images, with all sorts of OSes. 20:49:02 Made a few combos that were considered impossible by the OS vendors. 20:51:16 Before the age of virtualization, multi-boot was how booting multiple different OSes was done. 20:59:46 Back In The Day(tm) I was super-slick when I used Bochs to test if my experimental El Torito boot image would boot without having to waste time and another disc burning one and rebooting 21:00:02 though Bochs did it very very slowly it was still faster 21:28:20 Anybody familiar with raidz_expansion's on-disk format changes, if any? https://reviews.freebsd.org/D52174 21:38:47 cracauer: i piped the review to some folks that might know better 21:42:42 Thanks, kevans. 23:19:41 cracauer: sorry, but i really don't like the idea of permitting this feature (and i'm usually not a fan of complexity for something like that) 23:19:50 nothing personal 23:23:41 I think my proposed change to openzfs is better anyway, see review. 23:24:55 But that has a long turnaround. For 15-release we could restrict the installer to not set that flag. 23:25:00 right, but that only solves the problem for one narrow case 23:26:21 ideally we'd define a compatibility set (see zpool-features(7)) that one can use both in the installer and manually created pools that we guarantee will work and avoids problems like this 23:26:22 Yeah. With my install-on-second disk scheme I would be footshooting by default. 23:26:53 preferably you shoot your foot before the system has valuable data on it, because there's nothing to stop you from expanding that later if you provision a raid-z system like that 23:27:32 loader preventing it *is* the antifootgun, in this case 23:28:18 we have to explicitly OK features just to avoid weird situations like this where it kinda-sorta works but may not down the line, in ways that we don't have a good way of communicating or preventing 23:28:34 nothing stops someone from dual-booting linux off the same pool, for instance, which wouldn't be aware of our restrictions 23:30:44 features not being enabled are how you would more typically communicate that, and if they dual-boot, enable, expand without checking for compatibility, that's kind of on them 23:32:01 whats the feature in question? 23:32:23 raidz_expand 23:35:55 I don't get why all the drama? it doesn't change the ondisk layout of existing data at all 23:37:12 blocks aren' 23:37:14 t reflowd 23:37:41 as soon as you expand you're screwed, based on the theory 23:38:06 both during reflow and after completion because it moves to variable stripe width based on times 23:38:31 mm 23:38:43 why would we accept a feature on the root pool that bricks your system as soon as you try to exercise it? I don't like the risk at all, personally 23:39:21 raidz already has variable stripe width 23:39:42 a function of the # vdevs participating, no? 23:40:31 we already support this scheme 23:40:33 no within a raidz vdev 23:41:06 a small record will be on 1 datablock and a large one will be on as many datablocks as disks-Nparity 23:41:21 the description of time-based geometry sounds like post-reflow world differs from the pre-reflow world in a way that we can't handle 23:41:35 (right now) 23:44:23 that's not to say we couldn't, but our read implementation in loader is standalone w.r.t. to what's shipped by openzfs, things like different width math need to be ported over manually 23:45:42 ah ... right right. 23:47:12 i think i'd feel a bit different if it were only during reflow that we have some % chance of not being able to boot from it 23:48:11 reading through the section mentioned more , thanks for helping me get my brain in the right spot 23:49:55 otoh, this is all really easy to test by just popping it into the whitelist on a small VM and doing an expansion on a small boot pool