03:18:38 Dianora, LTNS. 03:19:13 so I have apache 2.4 with mod_php80 - what do I add for GD support in php? 04:11:25 CrtxReavr: would it not be "php80-gd", or is that too obvioous? 04:11:42 ghoti, I would think so. . . but. . 04:12:01 graphics/php80-gd wants to bring in lang/php80 05:41:32 lang/php80 should have the apache module or fastcgi 05:42:36 oh, i take that back, www/mod_php80 10:19:02 How hard it'd be, to implement Taiga on FreeBSD? https://taiga.io/ - checked entire community forum for "freebsd" keyword, no result at all. 10:19:04 Title: Taiga: Your opensource agile project management software 10:26:24 tercaL: you can try launching their docker image with podman. 10:28:46 kwiat: That was interesting idea! Definitely I should try. 10:29:42 tercaL: let us know how it goes :-) 10:30:18 kwiat: There are two different packages; https://www.freshports.org/sysutils/podman/ and https://www.freshports.org/sysutils/podman-suite/ 10:30:19 Title: FreshPorts -- sysutils/podman: Manage Pods, Containers and Container Images 10:30:22 Which one would you suggest? 10:30:53 seems to be written in python and js, so it probably works, but they don't have docs how to setup it, just their docker images to run 10:33:18 podman-suite is a metapackage and depends on podman, so just install that one, maybe you have use for the other dependencies as well 10:33:29 nimaje: Here they have "From source code", but looked like a horror to me :) https://docs.taiga.io/setup-production.html#_introduction 10:33:30 Title: Install Taiga in Production 10:33:52 Oh, thanks 10:45:57 Well, confusing..Would better try to run a Linux host on bhyve for that. 12:41:35 is there some magic trick to get amdgpu running that isn't in the manual? because when i try to start x i just get failed to load module amdgpu 13:47:54 FreeBSD as a daily driver, convince me yes or no. 13:50:32 hm, I use it as a daily driver for 5 years now I think 13:57:34 4 years here. 14:03:13 I recently quit HardenedBSD which is not FreeBSD. It was okay. FreeBSD will likely be easier than HardenedBSD. 14:03:27 i had been using it as my desktop daily for a couple years 14:03:58 I read an interesting essay about BSD compared to Linux. So far, I like the idea of having everything centralized, with ports for any additional software 14:14:50 Yammins: my first impressions are that freeBSD is a more cohesive experience, everything fits together nicely, but it also has less flexibility, compatibility is meh, and hardware support is iffy 14:15:34 skyenosaur: The hardware compatibility was what I was most curious about. I mean, I don't run any games on linux but I do use GPU for some computation. 14:15:49 so if you either have the freedom to choose your hardware by what works, or the ability and time to make what you have work, you'll be fine, otherwise you can be either lucky or unlucky 14:20:00 Yammins: mostly what works under linux works under freeBSD, but far as i can tell it takes a bit longer to arrive 14:27:11 So I have to repeat this: 14:27:17 23:19 < CrtxReavr> so I have apache 2.4 with mod_php80 - what do I add for GD support in php? 14:28:32 Obvious answer is graphics/php80-gd, but it wants to bring in lang/php80 which doesn't make sense. 15:50:28 Why not? 15:51:08 Because php already works with mod_php. 15:51:14 It's GD that's the issue. 15:53:21 So your issue is that you're not using the cli 15:53:48 That's not true for all php80-gd installs. 17:52:01 CrtxReavr, Maybe the path of least resistance is to let php80-gd bring in php80 and jet let it and then ignore it if you don't want it? 17:53:02 Maybe. . . though I seem to remember issues with that in the past. 17:54:49 I would build it up in a jail off to the side where it can't hurt anything and then try it all out there before dragging it into a production area. 18:56:07 ohhh my god i made it work. i have X now. i had to upgrade to 14 and install drm-515-kmod and now it works. that's certainly one way to get a freeBSD crash course 18:56:33 gmirror list shows me some of my mirrors as degraded. i have two questions: 18:57:24 1) gmirror list shows "Components: 2" but does not list the missing components, is this normal or did it "forget" about them? if this is normal, how do i show the missing components? -a doesn't seem to do the trick, neither for list, nor status. 18:57:46 2) what's the recommended action here? just "gmirror insert" them back in? 18:58:14 do you know why the components were removed? 19:00:35 RhodiumToad: it kept loosing one disk, i think it's a loose power cable, but haven't opened the machine yet. 19:01:21 it actually also lost the other disk once. both disks are on the same cable. and yes, i'm going to fix that. :P 19:02:05 if you trust it to be able to run both disks, then inserting the component again should do 19:04:05 will it rebuild from scratch, or recongize that last state on those re-added components? also, if the device paths are different (gpt/foo vs. ada3p2), will that make a difference? 19:04:17 (the latter is why i asked about showing the actual missing components) 19:04:43 it doesn't care about the device paths unless you used the option to store names in metadata 19:05:27 it should rebuild the whole of the newly inserted disk, I don't think it tracks dirty regions well enough to avoid that 19:05:38 oh, is that maybe the difference between my mirrors that show the gpt/foo paths instead of the adaXpY notation? 19:06:10 and if i want to go over gpt labels, in the future, should i always use that option when creating new gmirrors? 19:06:13 if you're not storing names in metadata, then which device name you see is somewhat arbitrary 19:06:31 since it'll depend on the order in which things were tasted 19:06:44 ah okay, so probably the same order that geom_eli asks for passphrases in– never been quite able to decipher the logic, but it seems to be consistent over boots. 19:07:07 yummy disks. 19:08:48 but okay, just looked it up, that option is '-h'. never used that, but i think i will from now on. hardcoded gpt labels should make things more robust, especially as i grow this setup and do more promiscious hotplugging stuff. 19:10:14 I haven't experimented enough with it. 19:11:09 i'll just try if it actually hardcodes gpt/foo then and if that's the case, mirrors should continue working and automatically re-sync, even when i rip out half the disks and switch them around. :) 19:11:45 phryk, Cautionary tale from me. I *killed* a SAS disk earlier this year due to a flaky power connector. If that is the problem don't put off fixing it. 19:12:45 rwp: thanks for the info, good point. it's just a lot of rummaging around right now, because i don't have a 19" rack yet and the server is below my desktop machine on top of which there's a lot of random shit. :F 19:13:26 Been there. Done that. Killed a disk. Bought a disk enclosure to rebuild upon and things have been rock solid since! 19:13:32 but yeah, all in all, not the most complicated operation, so i should probably just do it now, before i dive back into my python project… 19:13:45 ye, these are new disks, too. 19:13:48 I am pretty sure that when I have had gmirror'd disks that reconnecting the drives had them automatically detected and joined back into the mirror. I think. 19:13:58 2x 4TB WD RED. tripling system capacity. 19:14:42 rwp: yeah, but if a disk dies and is replaced with a new one, you'd at least have to re-create the partition scheme – earliest it can detect the thing is when a gpt label of the same name is created. 19:15:20 If the same disk is detached or powered down and then attached or powered up I think gmirror will detect it automatically and rejoin it into the mirror. 19:15:43 yeah, it should normally re-detect it based on the metadata on the disk 19:15:43 If a disk is replaced with a different disk then it will need to be manually partitioned and explicitly joined into the mirror. 19:15:56 rwp: yeah, that's the behavior i know and have come to expect. this is the first time i've seen it not do that, hence me asking here. :) 19:17:26 rwp: i'm not even sure if you have to insert it again if you use -h to use gpt/foo and then just create a new partition with gpt label "foo". i honestly think that would just automagically have it added in and synced becuase it's the same device path. 19:19:21 If this is in a deskside case then I assume you would need/want to power it down to mess with the power cables, ahem, in which case the reboot should have everything happy afterward when everything is there. 19:20:07 Also as far as using device /dev/gpt/labels instead of /dev/ada* names if this is a mirror then one can replace the parts one at a time in place and switch from one naming to the other. 19:20:55 right now, yes. but that's only cause i don't have a case with hdd hot plug cache. i'm already using 19" 4u atx cases for my machines because i'm definitely planning on a rack for this, so i should be able to get a fitting cage for this case. :) 19:20:58 Regarding your other question about listing out the missing devices, I think it doesn't actually have a way to do that. 19:21:21 The way I determine this is that I know the total list of devices that should be in the mirror and then list out the ones that are active and the rest are the missing ones. 19:21:23 that would be something i would have expected from both gmirror list and status, tho. 19:22:47 this is something that has been discussed at some length because of the implications for zfs 19:23:11 One thing I remember that caught me using gpt labels is that with gmirror it would not accept any labels containing a '-' character. It's not documented that I could find but I deduced that it would not work with a label with a dash. 19:23:30 in some respects it's a problem that gpt/blah and adaXpY are different providers rather than aliases for a single provider 19:24:25 rwp: huh? I've *always* used labels with '-' in them and never had any problems from that… 19:24:35 I also remember that gmirror forget has a -v option and I think when telling it to forget the missing components that at that time it will say what it is removing. And then those can be put back. 19:24:37 my scheme is literally - 19:25:23 I use gmirror with swap partitions and I could not get it to work with names like gpt/sw-3K1G3T97B which blew chunks until I switched to gpt/sw3K1G3T97B 19:25:24 like 'gpt/data-a' and 'gpt/data-b' make up the gmirror 'data' 19:25:49 blew chunks? what does that mean and how do i check for that?^^ 19:25:52 Maybe that was something fixed in 13? Looking at my notes those would have been from 12. 19:26:08 pretty sure i used this pattern since at least freebsd 11… 19:26:13 Sorry. "blew chunks" is an idiom meaning that it vomited. 19:26:40 ah okay, so if i don't see any obvious errors, i'm not having this behavior on my system. :P 19:27:34 curious. these mirror components didn't get attached by geli, it seems. when i re-attached all of them, one mirror automatically added the component back, but the others didn't… 19:28:29 My notes say it failed upon the gmirror insert action. I would need to set up an experiment to reproduce it now but my notes to myself are very clear that I could not get it to work for me. 19:29:18 weird. sounds like it must have been a difference in hardware then, but i can't really imagine hardware directly being involved with the gpt labels… 19:30:28 That does seem unlikely. 19:31:39 Anyway... If you are using a 4u atx cases then I have been able to add in 4x hot swap bays to those in the past. That's another option. 19:31:43 hardware shouldn't normally make any difference, except possibly in order of tasting 19:31:55 oh, maybe it was an ssd-specific problem. this fbsd14 now on my desktop is the first time i have a mirror over multiple ssds, and i think the ssd-accelerated stuff i got on my live server i setup during fbsd 12. 19:32:51 But it was on one of those that I had the Molex 4-pin power connector problem. Because it was too well used in my hardware hacking situation. 19:32:56 I bought a surplus 8x hot swap bay 2U rack case and that solved several problems for me all at one time. 19:33:54 okay anyhow, i'm taking your recommendation to heart and will replace the power cable in that machine now. and maybe the mirrors will just startup with all components again. :) 19:34:25 and i can also swap around the power led cables so that finally works, too^^ 20:01:20 mhh, problem – don't have another sata power cable for this psu. guess i can just lay it out in a way that's hopefully "better"… for some interpretation of that word.^^ 20:03:17 aand i found a spare psu with enough sata connectors. let's hope this works out. 20:30:26 "Component ada3p2 (device uboot) broken, skipping.", "Cannot add disk ada3p2 to uboot (error=22)." 20:30:28 mhhhh 20:31:25 uboot? 20:33:22 my /boot gmirror 20:36:22 oh. /dev/gpt isn't properly populated… 20:37:08 `gpart list ada2 | grep label` shows "uboot-b", but i have no /dev/gpt/uboot-b 20:46:16 `service devfs restart` doesn't fix that either. if i try inserting the other existing devices (data-a.eli for example) back into their mirrors, i get "gmirror: Not all disks connected" – does that refer to the missing disk or does that indicate that the disk i want to insert isn't connected? 🤔 20:55:00 okay, seems to refer to the "missing" disk. gmirror insert after gmirror forget on the mirror works. 20:57:08 try loading geom_mirror via /boor/loader.conf 20:57:20 or rc.local 20:57:26 basically load it late not in kernel 20:57:39 i have it in my loader.conf 20:58:00 ah I had to do that with a ZFS system where the boot drive was a gmirrored ufs partition 20:58:02 got all mirrors either fully up or syncing :) 20:58:04 wondered if it was the same thing 20:58:15 :-) 20:58:43 i don't think so. this only started acting up after it lost the drives while the system was running. multiple times.^^ 20:59:16 replaced the psu because i suspect one of the sata power cables. hopefully this is the last i'll hear of this. :P 21:02:10 one thing i could still use some advice on, tho – i have a bunch of firewall rules that depend on jails and vpn and those don't work before i do a manual `service pf reload`. what's the clean way of automating this away? 21:23:15 man pf.conf claims you can put parentheses around interface names and the rules will be updated whenever the address(es) on an interface change. i already tried that in like half a dozen different ways and the only thing it ever gives me is a syntax error… 21:32:49 wait, i think i know what's going on. it's saying "parentheses" – () – but really means curly braces – {} m) 21:32:59 will see at next boot whether that was it. 22:18:14 is there anyone who can edit this wiki page? desktop mostly works – even appears to have glx – but the new stable version that’s not officially released somehow has NO browser in its binary repos except for fucking luakit, and there’s the small problem where i can’t actually access my home drive at all because it’s LUKS encrypted 22:18:26 oops wrong paste sorry im stupid 22:18:39 https://wiki.freebsd.org/Graphics/AMD-GPU-Matrix 22:18:40 Title: Graphics/AMD-GPU-Matrix - FreeBSD Wiki 22:18:57 because my video card is not in there but it works 22:22:44 well, it works on 14, that is, not on 13.2 22:29:01 skyenosaur: rx6...? 22:32:43 phryk: yeah rx 6500 xt, it’s navi 24 and not navi 22 like the 6700 that is in the list 22:34:26 skyenosaur: i think that's similar my rx6600xt. there is a recent fix for a memory leak which i'm not sure is in the official port/package yet and in the issue the person doing the fixing mentioned stumbling over another bug during that and wanted to fix that too before a new release of drm-515-kmod. 22:35:43 so, you might want to check if your setup has the memory leak. vmstat -m | grep lkpikmalloc the size that gives you should probably be less than 10MB, definitely way less than 100MB. 22:35:53 phryk: ahhh i see! that makes a lot of sense 22:36:05 see also https://github.com/freebsd/drm-kmod/issues/258 22:36:07 Title: drm-515-kmod amdgpu memory leak · Issue #258 · freebsd/drm-kmod · GitHub 22:36:08 258 – The dialog_menu function's scroll saving feature is broken. https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=258 22:37:50 phryk: seems to be around 10 MB, but i also don’t really have anything running at the moment 22:38:35 i was just relieved to have X after a half day of fighting :D really fun way to get started on a new OS 22:38:51 skyenosaur: the leak occured as soon as i had a running X session and seemed to always leak at the same pace (about 20KB/s). so if your machine has been running for a couple hours and it's still at ~10M you shouldn't be affected. 22:39:38 phryk: yeah its been on a while, thank you for the heads up! but i think i’m safe from this one 22:39:49 good. :) 22:40:30 mhh, freebsd doesn't have SSO for all its services, does it? i.e. i can't use my bugs.freebsd.org login to edit that page? 22:47:54 just confirmed – it got upstreamed into ports by now. :) 22:50:47 phryk: haha i just turned on glxgears to see if activity would actually make it leak and it does! it gets biggerrr. i did install the binary package though, so if the fix is in ports i should be able to install that one 22:51:06 phryk: we do, but only for freebsd accounts i'm afraid 22:51:37 well, doesn't apply to the wiki 22:51:54 kevans: "freebsd accounts"? so you can log onto my machines? :P 22:55:43 skyenosaur: does it get smaller again when you close glxgears? 23:12:47 phryk: no 23:13:12 then maybe you do have the leak^^ 23:13:55 it also gets bigger when i do lots of movements, switch windows etc. i think it really was just so small because i have done *nothing* on the pc 23:15:11 but oh well, i just typed portmaster -a and then hopefully this will also update the offending kernel module... i suppose i’ll know tomorrow 23:15:20 good luck. :) 23:15:51 thank you! 23:19:23 also really appreciate that warning because i would have been caught SO off guard by that inevitable OOM kill :D 23:20:19 ye, this was one of multiple confusing things i had to deal with setting up my new system. thankfully RhodiumToad and a couple others were super helpful about triaging this. :) 23:24:29 well, i did install FreeBSD looking for a challenge, it's only right that it should start like this, it's fun 23:26:12 i waited about a year to actually get driver support for my new gpu and when i finally updated to 14, my system disk died. then when i upgraded to the new board/cpu i got some more problems. this is the system i use for work and that was a bit less of a fun experience. ^^; 23:43:07 phryk: haha that somehow sounds even worse than the day i had. i tried on 3 different machines. one ex-chromebook where the custom efi messes everything up, one laptop that worked right until the HDD died 23:43:44 and then this thing where i got stuck with a video card that didn't work. i mean my whole idea is to start making things work that do not work yet but maybe not on my first day 23:44:24 /join jellyfin