00:25:10 well, you can propose the patches upstream already, upstream and the port don't need to way on one another there, in the port you want to make the current release build on freebsd and upstream you want to ensure that the next version builds on freebsd, there are no dependencies between those two goals 03:12:13 The multitail program in FreeBSD has a memory leak. If I leave it running it eventually absorbs gigs. It doesn't have this leak on GNU/Linux systems. I guess I should debug it at some point. 03:15:29 Bugzilla time! 03:19:14 I am embarrassed to say I simply rebooted the system a few times before chasing down that it was my multitail run in a terminal that was the program eating up gigs of ram, which forced a growing swap usage. 03:20:12 And now it is pretty easy to 'q' and then C-p Enter (one could !! it) to restart it so that's what I have been doing since I figured it out to this program. 03:23:44 rwp: The terminal is running away with mem? Terminal emulator? 03:24:12 XTerm is okay. It's multitail. 03:25:11 Oh, I see. Still the same problem existing, you're just saying it took some time to figure out it was, in fact, multitail causing it. 03:26:04 Right. Check out how much swap it consumes when it is deep into the leak: https://www.proulx.com/tmp/multitail-memory-leak.png 03:30:57 what's wrong with `tail -f *`? that doesn't leak 03:43:23 multitail runs tail -F across multiple files and displays all of the files in one terminal window, optionally merging output of multiple files. 03:45:41 Unfortunately for me there are cohorts who have created literally tens of apache *access.log files and I can use multitail --merge-all *access.log to merge them all in one window to look visually for abuse patterns. That's a useful feature. And can look at multiple other files for abuse such as multitail auth.log fail2ban.log messages to view multiple logs in different curses windows. 03:47:00 And I suggest looking at the difference between tail -f and tail -F. I am bummed that the behavior of -F didn't make it into -f as it ought to have done. But that's spilled milk under the bridge decades ago now. Always use -F instead of -f if there is any possibility that the log file will be rotated. 05:22:17 how to fix it? 05:22:17 # jexec a xfe 05:22:18 FXApp::openDisplay: unable to open display localhost:10.0 05:27:43 Chip1972, You appear to be trying to run a GUI application inside of a jail without connecting that up with the desktop outside the jail. 05:33:29 Chip1972: what is your goal 05:37:02 i want to learn 05:37:12 xfe is a gui file manager. Probably not a good choice to be trying to use in a jail. ls is probably a better choice for use into jails. 05:38:37 what? 05:39:15 jail != jails ? 05:39:47 I see the "jexec" followed by "a" and I think you are trying to execute a program in the jail named "a". And the command I think you are trying to run is xfe a graphical file manager. 05:40:23 if you want to learn then either try it without a jail or read up on the x protocol to learn why it's not working 05:40:24 But if you are just learning then attaching the graphics from a jail to the hosting desktop is rather involved. There are lots of articles on how to do this on the net. I quoted one earlier today. 05:40:58 Better while learning to stick to the simple command line in a text terminal. That's the best way to learn. 05:43:50 I am scrooled 06:52:01 ivy: hey what's up? 06:52:42 can you give me some info regarding the process of contributing to the freebsd source? 06:57:52 https://docs.freebsd.org/en/articles/contributing/ 07:55:49 no I mean how does the hiearchy thing works? 07:56:05 like who gets to be the commiter and maintainer this and that? 08:08:50 "maintainers" only really exist in ports, anyone can become a ports maintainer by submitting a new port or taking over an existing unmaintained port. commit access (to either src or ports) is usually given to people who have a history of submitting useful changes 08:10:55 ivy: okay who gets to decide if someone is a commiter? and are you one? 08:11:35 there are mentors and mentees 08:11:59 the book i am reading tells me to find a commiter to talk to before trying to contribute 08:12:02 anyone can propose someone for commit access, it's decided by a vote (of core, i think... or is it srcmgr? one or the other). i am a committer, yes. but... being a committer doesn't really mean anything, it just makes it a bit easier to get your changes into the code 08:13:40 okay alright 08:14:15 what is a "port"? is it part of the system image shipped online? 08:14:30 they're third-party packages, i.e firefox 08:14:46 okay alright 08:15:11 thx for the tips 08:16:07 do you guys have a conference like debcon every year, where people show off the latest cool stuff? 08:16:20 yes, bsdcon etc. 08:16:39 freebsd.org lists some of the events 08:17:06 okay alright 09:35:05 rwp: interesting, I am the packager for multitail. I'm actively using it in two separate virtual TTYs with screen with tailing logs and tailing script execution output, haven't noticed a high memory consumption — although I have 64GB ram on the server so didn't have to watch for any OOM. if you can reproduce it somehow (maybe feeding huge amounts of logs and seeing if RES raises significantly), we 09:35:11 could get a hint of what happens. that said, the packaged git tag is quite old and there have been a couple changes added into master ever since: https://github.com/folkertvanheusden/multitail/compare/247e7ff727d9a8bed410f1bbf86f247c059546d9..master 11:47:11 zi: why does net/fiche create a user when the service doesn't use it and runs as root? did you just forgot : ${fiche_user:=fiche} in the service? 12:04:39 hm, the pidfile has to be created before too, so that daemon can write it, but seems to work fine like that 12:18:26 pkgbase-based upgrades are fast 12:19:47 basifying 14.3 and transitioning from 14.3 to 15.0 took only 20 minutes and one reboot 12:57:37 bleh, kinda hate when I find something I need and their only supported installation method is.... docker compose 12:58:29 I know FreeBSD has containerd & runj but haven't looked to see if there's any reasonable way to do something like docker-compose with this, or if I should just give up and run some minimal Linux inside bhyve 13:06:41 I see nerdctl supports this, but is also pretty early in terms of FreeBSD support, hmmm 13:20:29 well, if docker is the only supported installation method, then it hints at the installation being so convoluted and shitty that they wanted to sweep it under the carpet 13:21:11 well yes obviously but people get pissy when I tell them this 13:36:49 I've made the point repeatedly to people: having the *option* of Docker is great, when deploying and integrating software, or even just running software whose dependencies, required configuration, or privileges just badly overstep. Developing software under the assumption that it will be there is not. 13:54:08 hodapp: it will probably work at least under two different OSes, WSL is proven to support Docker images 13:55:03 so no worries, you can change OS, and if you want to really run it under FreeBSD controll, then wirtualise the OS hostion your app 15:14:55 hodapp what software is it? whatever it is, it has to run without Docker anyway :-)) 15:15:23 I really wanna know :D just to add that software to my blacklist ahaahah 15:31:19 antranigv: immich 15:31:43 antranigv: someone's made derivations in NixOS for it, so it definitely *can* run without Docker, but I dunno what's involved 15:32:20 also: https://github.com/immich-app/immich/discussions/1657 18:51:24 mzar: my 14->15 pkg base upgrde went very very wrong 18:51:28 but i just yolod it 18:52:33 hodapp: `podman compose up -d` 18:52:51 after `kldload linux` 18:53:48 i don't know if immich would ru nit it 19:07:34 rtprio: it requires adding signatures from 15.0 and yes, I had to replace master.password and group with old verions 19:08:16 but the whole transition is seamless if you have root on ZFS and the ablility to create new BE 19:10:02 it was my first try to pkgbasify and do major version upgrade, and I am impressed <- Good Job PkgBase Maintainers ! 19:10:38 I will script it and make transition of a bunch of workstations next month 19:24:00 even in the 14.1-14.3 the /etc/rc.d/ldconfig never properly merges 19:56:49 karolyi, Hi! Just replying that I see your messages to me. No immediate time but I will see what I can do to produce a reproducer for you in the near future. :-) 20:00:47 rwp: whenever you do reproduce a reproducer by reproducing the reproducable, please open a bug ang ping me there :) 20:01:09 *and 20:05:33 Feel free to break out into a Gilbert and Sullivan Pirates of Penzance song but I am not joining in! Last time I sang for my supper they called it extortion. :-) 20:22:06 Say, is the latest version of FreeBSD for armv6 raspberry pi 13.5-RELEASE because that's the newest version that'll build or is it because it's the newest version someone's bothered to build? 20:43:42 zip: armv6 is no longer supported 20:44:45 Nor is i386 :-( 20:45:51 i386 should be 20:48:18 I think 15.0-release might be broken right out of the box if you pick the pkgbase install option. I do see there is a bugzilla that seems to be actively commented on https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=292214 20:48:55 15.0 aarch64 to be clear. Since that seems to make a difference. 20:49:20 i think i386 is dead in 15 20:49:36 Ah, so no point attempting a 14 build then 20:50:39 Once again I should obey the note I stuck on the damn thing that says "get a real computer to tinker with" 20:50:39 AmyMalik: could be true, but for 14 i386 is still supported 20:51:27 zip: armv7 is still among supported architectures 20:51:34 the first Pi is arm6 i believe 20:51:52 mzar, I believe I once golfed the i386 kernel down to be able to boot a VM in like 81 megs of ram 20:52:02 RPi2 is v7-based 20:52:55 the bootloader prevented me from going any further, I believe, had I used grub I might've been able to get it up and talking on 64 20:53:49 mzar, yes, the last version with i386 is 14. 20:54:53 OK 20:55:35 Yeah. I should put the zero in the bin 20:55:49 It was shit when it was new and that was over a decade ago 20:56:28 I just hate wasting old hardware and buying new hardware but I guess the solution to that is to cycle out the pi 4 from its home assistant duties and then tinker on that 20:57:07 I think it's pretty neat there's computers so light and low power that you could use one as a computation sidecar to your phone or iPad if you needed to 20:57:21 zip, or use a diffent flavour of BSD, and even Void Linux? 20:58:41 But realistically I also solved that problem in 2019 when I didn't have a computer for a while with a cloud-init script and a couple shortcuts and would boot up a VM for a penny an hour if I needed Python 21:03:39 i wonder if i have or could develop the skills to re-port FreeBSD to the 686 (I believe the i386 port was for 686 and up), solely as a one-off entertainment vehicle 21:09:26 ?! 21:09:35 hi!, anyone using zelta for zfs backups here?. I'm using it since a couple of months now and it works great. Now I'm looking for a way to delete old snapshots, I don't know if Zelta has a built-in feature for this. 21:10:48 AmyMalik: FreeBSD was running fine on 486, without issues IIRC 21:13:26 ah 21:19:32 doesn't netbsd still support many of the older archs? 21:20:24 i believe so 21:20:55 I ran it on an a2000 amiga with a 120 GVP scsi hd years and years ago 21:21:15 have never tried it on an intel based system 21:22:24 did not really do much with it except conclude it worked fine 21:23:11 I had a 68040 card in I think 21:23:26 not particularly fast