00:00:22 Howdy all! 00:01:27 Are you actually a 486? Impressive in this world of 3MB ls(1) 00:02:23 Right now, no. I'm on my FreeBSD machine. I use ircjr on FreeDOS from time to time to log into irc though. 00:02:30 Ah 00:03:02 meena: do you happen to know what the import format is for gts 00:03:06 Yeah, that PC has a 486DX 66MHz, 16MB RAM, ~500MB SSD (CF card). It's a great Doom machine haha. 00:03:38 I was trying to get mere NetBSD MINIMAL booting on a 12MB DECpc 486. No dice. Just needed to image some SCSI disks. 00:03:47 But also have networking 00:04:41 (in service of restoration of a NeXT '030 cube) 00:04:50 Yeah, I looked into doing something similar but even NetBSD is just too resource hungry :/ Plus any modern version of X doesn't support video adaptors over the ISA bus :/ 00:05:00 Oh, X was 100% out of the question lolol 00:05:30 I wanted to ssh blah dd of=/dev/sd1 00:05:41 I wonder how Xenix would run on the thing lol 00:05:51 Probably fine, but idk if xenix has NE2000 drivers 00:05:59 And definitely no ssh 00:06:10 Yeah, I need NE2000 drivers for my Realtek NIC. 00:06:28 Didn't know realtek had an ne2k nic, interesting 00:06:32 I couldn't get the Realtek drivers to work but luckly it's NE2000 compatable so I use that instead. 00:06:37 Actually nvm this wasn't ne2k, it was an actual DEC tulip 00:06:45 Like, from digital 00:06:58 hah, i bought three ne2000 cards for $.01 and 2 of them worked 00:07:12 I have one that I keep on the shelf just in case 00:07:13 what a day that was, built possibly the slowest router i could at the time 00:07:19 Hahaha 00:08:34 That sounds like torture. 00:08:41 I used it as a router back in the day when cable modems only ever had a 10Mb PHY 00:08:49 How far we've come 00:09:36 ...I still can use a dialup modem if I need to (which I've used a few weeks ago for fun). I'm one of the last few lucky people to have a real POTS line to my house. 00:10:16 I like to dial into BBSes on my 486 from time to time too. 00:10:23 Lucky. I just have a grandstream ATA and a bunch of western electric phones behind a ring regenerator. I've been working with GS support to get fixes in for supervision behind the booster. My IAXy never had a problem. 00:10:45 Their debounce time on the codec is too long, we've found 00:11:19 I'm about 700 or so feet from the distribution building so I get like the fastest VDSL (bonded) connection you can get. No fiber here yet. 00:11:59 Maybe I'm the lucky one, lol. We were one of the first fios markets and after having it for 17 years I have to say it's a good service, even if by a truly evil company 00:12:29 Who is pulling out all the copper because copper is still a regulated service around here 00:12:44 lol well I'm with Windstream but yeah, I think Verizon is worse (as a company). 00:13:10 Without question. But their product isn't oversubscribed garbage like their HFC competitors here 00:13:26 I mean I get about 150 (up to 200) MBs down and around 50MB up. Not great but can't complain for living in a town of 230 some people. 00:13:28 400/10 service. You can just about send the ACKs out fast enough to get 400 00:14:14 I do miss my 100/100 fiber from the other place I used to live at but I like telephony so it's a trade off. 00:14:42 That is fair. I'd trade my connection for 200/50 copper if it were offered 00:14:50 I restore old phones, lol 00:15:53 Part of the phreaknet? 00:16:27 I am not, have thought about it on and off for years. I probably should considering I listen to Evan Doorbell tapes when I walk 00:17:44 I'm not familar with that. 00:18:04 Hey, I've got to get going. I may rejoin later though. 00:18:13 You're in for a treat http://www.evan-doorbell.com/production/group1.htm 00:18:14 Title: Evan Doorbell's Telephone Tapes - Group 1 Playlist 00:18:17 Take care! 01:32:35 hi, noob here. i'm on 14-current and just ran pkg-upgrade, now i can't use git anymore. the error i get is from ld-elf.so.1 about an undefined symbol. i've been searching around but mostly found issues that had nothing to do with my particular one (stuff about jails and such). Any pointers? 01:38:09 did you run this in a jail or your base system? 01:38:17 base system 01:46:22 i wish i had something more constructive to say than a noob probably shouldn't be running on 14-current. but i'm thinking 01:47:01 well i did try, but 13 stable left me without a bootloader after 6 different installations 01:47:14 i want to say that this can happen if you wait to long to update the base system and then you try to get an updated pkg repo and use newer binaries than were intended 01:47:26 you probably want release 01:47:32 13-release 01:47:54 will the system respond to freebsd-update fetch and then freebsd-update install? 01:49:27 fetch tells me that it can't upgrade from something that isn't a release and that i can either upgrade from source or upgrade to a release version prior to running freebsd-update 01:50:17 haven't tried install 01:52:01 yeah, i don't know the proper way to update a current system. lemme try to look real quick 01:56:00 i can't figure out what you should do. seems like git is essential to working with the development branches (stable and current), so if you can't revert what you've done, then you'll have to see if someone will walk you through it, or you should just start with 13.1-release 01:56:02 https://docs.freebsd.org/en/books/handbook/cutting-edge/ 01:56:04 Title: Chapter 25. Updating and Upgrading FreeBSD | FreeBSD Documentation Portal 01:56:31 wish i could help more. i've encountered that error before, but never with the base system 01:57:02 thank you, this is still helpful to me 01:57:13 and i usually ended up creating a new jail and migrating the data over. wish i knew how to fix it, so i'd be curious if you figure it out or learn how 02:02:06 deadguy: is it __fbsd__something 02:02:43 if it is, you need to buildworld first 02:02:46 rtprio: yeah, that's the one 02:07:10 i closed the wrong pane... i'll look into buildworld 02:20:10 deadguy: OR you can rebuild them from ports instead 02:20:29 is there a way to run "make install" as an ordinary user and have the owner and permission or maybe just own information added to an mtree file? 02:21:17 i don't really have any ports installed. i've been getting everything through pkg. i was maybe looking into downgrading this particular package (git) but figured that's probably not the best idea 02:22:00 ports are packages 02:22:16 rebuild it from source, or rebuild the world/kernel 02:23:17 or rather, ports also install packages 02:39:06 is zfs faster than ufs? 02:39:13 given a i7 8g ram 02:39:16 desktop 02:39:24 8200 rating on cpu bechmark 02:39:41 right now keeping simple on these usb3 drive 02:39:44 ufs 02:39:48 nothing raidded 02:58:01 concrete_houses: too many variables to give you an answer 03:14:30 crb: i don't think so; what are you trying to do 03:16:30 rtprio, I am building images for an embedded board. There are tools like makefs and mkimg that allow you to build disks images based on a filesystem, but it seems your source filesystem has to already have the files with the right owner and permission 03:17:30 rtprio: if you could do a make install where the owner was encoded in an mtree then it would seem you wouldn't have to be root to do an install as I think you could pass that to makers 03:18:24 perhaps a small enhancement to install so that you pass an mtree file and it install the file with your ownership but appends the true final owner to the mtree file later? 03:18:44 so is the problem putting it into the right place /tmp/myroot or knowing the uid/gid/usernames you need when you're done 03:20:13 if you're doing this in a jail and then making an image you could `make DESTDIR=/myroot install` but that only partially solves your problem 03:26:24 rtprio: when doing make install the installed files need to be owned by a particular user, often root. I think the pretty much forces you to run make install as root even if you are installing to a private directory that you own. I think you'd have the same issue if you ran in a jail. The install would need to run as root in the jail. That's al little safer but you still need root to setup the jail right? If there were a way 03:26:25 to write all the install files as my user ID and encode the true owner in an mtree file I could perhaps do the install myself and then give the more to makefs for when it build the filesystem 03:34:04 i see 03:38:10 with using ports i dont see a way 03:40:23 short of making a jail and using that to build a mtree 03:40:45 but you still need several root commands 03:47:56 rtprio: yes but it seems a small enhancement to install where install writes the file with the users ID but writes an mtree entry with the correct final owner and permission might work 03:50:43 i think youll find thats a signficant modification to the ports system 03:51:42 i dont get why you cant do it as root 05:13:15 When I do `pkg search/install Xvfb` it shows nothing but the package is there https://man.freebsd.org/cgi/man.cgi?query=Xvfb&sektion=1&manpath=freebsd-release-ports 05:13:16 Title: Xvfb(1) 05:17:49 The same is true of 'pokerth' and I don't know why: 05:17:51 https://ports.freebsd.org/cgi/ports.cgi?query=pokerth&stype=all 05:17:52 Title: FreeBSD Ports Search 05:18:11 rtprio: sorry not the ports system, I mean building a kernel and world, I should have been using make installkernel and make buildkernel 05:19:07 actually what I think I'm looking for is at the bottom of this page: http://bsdimp.blogspot.com/2020/06/ 05:19:08 Title: Warner's Random Hacking Blog: 202006 05:20:54 -DNO_ROOT seems to enable it 05:27:33 New to freebsd, my laptop heats up as soon as the boot process starts and waits for my passpharse to decrypt. what could be the reason? This never happend with my linux installs (not complaining, just providing context).any ideas? 06:10:59 guess the problem should be caused by the bootloader, boot0/boot-easy. I need to figure out if using grub solves that problem 06:37:49 for i in {9001..9099}; do echo $i ; done ; -> for: Command not found.. i: Undefined variable. 06:38:40 Letiute: man $SHELL and search for for loops 06:39:54 sounds more like a job for 'until' loops 06:40:28 maybe a pkg install bash problem 06:43:41 bash solved it 06:51:55 anybody else using unifi7 ? 06:54:50 how long is a piece of string ? 06:55:47 at least one char if it's a valid C string 07:01:25 sozuba: I'd guess the initial powersaving level of the cpu hasn't been set that early in boot. worth filing a bug at https://bugs.freebsd.org/ 07:01:28 Title: FreeBSD Bugzilla Main Page 07:03:31 Letiute: xvfb is not a port of and in itself, it is part of xorg-server 07:05:11 xtile: pokerth is its own port, but is BROKEN `error: use of undeclared identifier '_1' and doesn't build 07:05:53 both these bits of info are in their respective Makefiles and dirs 07:06:07 aha 07:06:13 thank you 07:06:25 https://cgit.freebsd.org/ports/tree/games/pokerth/Makefile#n25 07:06:26 Title: Makefile « pokerth « games - ports - FreeBSD ports tree 07:06:38 many ports secrets are listed https://docs.freebsd.org/en/books/porters-handbook/book/#dads-noinstall 07:06:41 Title: FreeBSD Porter's Handbook | FreeBSD Documentation Portal 07:06:50 explaining broken 07:06:57 thanks :D 07:07:14 yw 07:20:16 dch so how do I install it? I can install it in linux though 07:21:12 rtprio magic of googling, irc is that you do NOT have to read and search ALL the man pages. Me simply switching to 'bash' solved my problem. I do understand 07:21:13 Letiute: its part of x11/xorg-minimal I believe 07:21:31 dch so its installed already? 07:22:33 honestly I don't know, you can ask in #freebsd-xorg for a qualified answer 07:23:51 i have xorg (full install) installed and don't have xvfb 07:24:07 xtile: what do you mean, "don't have" 07:24:17 i.e. what are you expecting to find but don't 07:25:06 Letiute: i'm not sure you do 07:25:47 dch: don't have the xvfb.1 man page Letiute referenced. (i have no idea what xvfb is or does) 07:25:59 virtual frame buffer, iirc 07:26:36 you run it, set DISPLAY to it, run shit, capture the screen. that sort of thing 07:26:40 * xtile nods 07:26:41 https://man.freebsd.org/cgi/man.cgi?query=Xvfb&sektion=1&manpath=freebsd-release-ports 07:26:42 Title: Xvfb(1) 07:26:52 aah so you are looking for the Xvfb binary? 07:27:04 dch trying to find a way to install it 07:27:11 yeah, all i meant was that it's not in xorg or xorg-minimal ;3 07:27:12 pkg dont have it 07:27:34 sudo pkg install -r FreeBSD xorg-vfbserver 07:27:47 its not entirely obvious how to figure this out 07:28:15 https://www.freshports.org/ports-mgmt/pkg-provides/ 07:28:16 Title: FreshPorts -- ports-mgmt/pkg-provides: Pkg plugin for querying which package provides a particular file 07:28:21 dch I heard that pkg is known to break xorg 07:28:25 I grepped /usr/ports for -i Xvfb 07:28:28 not sure if its true 07:28:33 have heard no such thing 07:28:43 rtprio I am sure you have not. 07:29:02 Letiute: possible. but given the enormous number of people using xorg on FreeBSD daily it is probably going to be a very visible bug 07:29:05 I would notice 07:29:23 anyway I found x11-servers/xorg-server/pkg-descr-xvfb 07:29:44 and x11/xorg/Makefile 07:29:51 which has Xvfb:x11-servers/xorg-server@xvfb 07:29:58 the @ tells me its a flavour 07:30:06 so install x11-servers/xorg-server@xvfb 07:30:29 and now I have Xvfb 07:30:30 binary 07:31:48 I use FreeBSD as my desktop and pkg installs xorg fine. 07:34:22 where is my oss audio gone 07:50:11 dch ok 07:50:28 so I want " xorg-vfbserver-21.1.4_1,1 " ? 08:01:38 Letiute: dont't complicate it 08:01:48 sudo pkg install -r FreeBSD xorg-vfbserver 08:01:51 ^ thats whats you want 08:02:19 pkg(8) will do the right thing 08:03:21 otis: that last unifi upgrade was a pain. lucky I read UPDATING 08:04:01 I couldn't get it to load the monthly auto-backups so it was a zfs rollback to december and make a manual one then 08:04:08 not like I make frequent changes 08:04:30 dch whats the difference between pkg install -r FreeBSD xorg-vfbserver and pkg install xorg-vfbbserver ? 08:04:43 dch: what was the problem? should the UPDATING entry be more verbose? 08:05:04 otis: no I just wanted to say thanks for the entry! 08:05:33 I have no idea why unifi rejected the autobackups 08:06:55 Letiute: you can man pkg(8) for details, but `-r dch: it worked flawlessly for me, i did it on more than one installations. 08:08:15 anyway I still have one left. 08:13:39 dch isnt the default repo "FreeBSD' already? 11:01:55 you won't believe this 11:04:17 and we won't not believe it either if we don't know what it is 11:13:22 it's about that stupid power virus again. It was, of all things, PCIe. By running devctl disable on some PCIe devices I'm sure I won't be using any time soon (thunderbolt), I was able to bring idle power consumption right down to within a margin of error from the competition. 11:14:16 To be clear, I'm not recommending anyone do this, just reporting my observations. (I guess I should do a writeup on my blog) 13:22:55 AmyMalik: also a bug report 13:24:47 that makes me wonder if https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269823 goes deeper than virtio 13:24:49 Title: 269823 – rand_harvest produces 100%CPU on 1 CPU with virtio_random.ko loaded 13:25:35 how do I properly set XDG_RUNTIME_DIR “ 13:25:45 for sway? 13:27:46 joes: depends on how you start sway 13:28:16 Is there a way to get package to install/upgrade to simply ignore the dependencies on postgresql? I want a specific version (15.2) but theres also like 4 packages that want to install different versions (some 14, some 13) and overwrite my choice. How do I resolve this? 13:28:21 it has cometo my attention that my use of the term "power virus" is misleading. 13:28:40 pkg* 13:32:16 meena, that one is definitely unrelated 13:33:14 Would be great if pkg supported some paramater that specified "allowed alternative dependencies" where for a specific dependency being requested for a pkg other alternatives in list (if installed) will suffice 13:33:16 AmyMalik: well it's PCIe 13:33:32 mh 13:34:13 adonis: yeah, I don't think we support that (yet). 13:34:32 I'll do the blog writeup and then if it turns out to be PR worthy i'll do a PR 13:35:51 I err on the side of creating bug reports: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=269906 13:35:53 Title: 269906 – dtrace requires empty expression to print anything from matching search 13:36:53 meena: something like I am mentioning above is in the works? 13:38:28 adonis: somebody in #bsdports on EFnet should know. maybe there's a wiki page or a review 13:38:53 i dunno. my interests in ports are limited, so i don't keep up with all developments 13:46:58 meena: from a virtual console, not from a manager 13:47:46 joes: so it should be enough to export it in your shell 13:49:18 joes: i have this in my .bashrc https://bpa.st/N725A 13:49:19 Title: View paste N725A 13:50:37 angry_vincent: thanks. Sucks to do manually 13:53:20 doing the writeup is a way to organize my thoughts and findings into something good quality enough for the bug reports I often see 13:54:12 joes: since you starting sway manually anyway, i find no issues with that. it is set automatically once you have your shell up. 13:54:56 Is it possible to have a binary update mechanism for -CURRENT? 13:55:39 joes: another way is to set it up from within .login_conf 13:56:54 Whats the best way to formate a new disk? gpart? 13:57:51 angry_vincent: I might have overwritten $HOME for dual-booting. Is there no /etc/skel? 13:59:57 joes: this is something that exist on linux distributions, i think 14:01:31 yes. Is a new ~ not populated? 14:05:32 no idea what you mean 14:06:01 does a new user get an empty homedir? 14:09:28 adonis: best is to setup your own poudriere repo 14:11:34 joes, normally not. A new user's home is populated with a copy of /usr/share/skel/ (with all filenames changed, s/^dot//) 14:43:08 V_PauAmma_V: ah, that’s where it comes from 16:44:12 antranigv: Im testing jailer out... Not sure if Im doing it wrong, jailer create -b bridge0 -a 10.0.0.2 efs_dev_db <- should attach to bridge interface and set ip to 10.0.0.2? right 17:18:37 what is the equivalent of linux-kvm on freebsd 17:20:13 gzar: bhyve? 17:20:52 so i cant really use qemu with kvm 17:21:00 or other kernel-level virtualization 17:21:27 gzar: What's your specific goal? 17:21:35 antranigv: aah. found my mistake 17:21:52 running a linux virtual machine for some software. i'd like to have some form of acceleration 17:22:57 gzar: https://wiki.freebsd.org/qemu 17:22:58 Title: qemu - FreeBSD Wiki 17:24:31 I wonder if qemu is the right answer there. "I want a VM" seems reasonably well-solved by Bhyve or VirtualBox. 17:24:58 I dunno.. he specificly asked for qemu :-P 17:25:22 https://en.wikipedia.org/wiki/XY_problem 17:25:23 Title: XY problem - Wikipedia 17:27:45 qemu is a front-end for a lot of virtualization, similar to how libvirt is - and libvirt includes support for bhyve 17:29:10 Ah, and that reminds me that I didn't mention Xen. It's a sad day. 17:37:29 qemu as a front-end for bhyve requires that someone interested in doing the work does it. 17:44:15 i was asking about qemu cause linux has a module for it called kvm and netbsd has a module for it called nvmm, i assumed freebsd had something too but any other virtualization software with hardware acceleration will do 17:46:53 many of us here aren't fluent in Linux (any more) 17:47:21 last time I worked with Linux is probably pre-pandemic 17:49:13 Hi. Does anyone use wayland+sway out here who knows how to get GUI applications under Jails working with Wayland? Most documentation and support online only Focuses on X11 + "xauth +" .. I want to get things like chromium / firefox work under a jail with wayland 17:50:05 https://forums.freebsd.org/threads/running-google-chrome-in-a-dedicated-linux-jail.85491/ ... Like this guide is working for a lot of people but focuses on X11. But I use wayland 17:50:06 Title: Running Google Chrome in a dedicated Linux-Jail | The FreeBSD Forums 17:50:21 I always get the "Cannot open display" error with wayland 17:53:35 its fine, i just have to get this linux thing running and i'd prefer some accel. i'll try bhyve 18:47:55 antranigv: nice work with jailer... =) trying out some use cases i have. havent checked out all the features yet - is there a way of pass along some custom config lines when using snapshots? 20:36:19 drobban thanks! I will be "freezing" the code and doing only bug fixes. Jailer 2 will be rewritten from scratch with better features. 21:15:21 gpart was showing my hdd yesterday. But I wonder why `gpart show` is not showing my hdds. Other ssd,nvme are shown.. One of the hdd is even in a pool and working fine. also shown in /dev 21:33:16 gpart does not show "hdds", it shows partitions 21:33:53 if you didn't create any, it won't be there in gpart output 21:34:40 yuripv I see. How can I see disks then? 21:35:11 yuripv by the way I do have /dev/da0 and /dev/da0.eli 21:35:35 maybe .eli is not a partition? 21:38:01 diconnected 21:38:07 yuripv by the way I do have /dev/da0 and /dev/da0.eli 21:38:10 Letiute, Try: camcontrol devlist 21:38:10 maybe .eli is not a partition? 21:38:17 rwp ok 21:38:54 "camcontrol devlist" will show the devices that are attached at the device level. 21:39:03 Also try: sysctl -n kern.disks 21:39:22 That will show the device names the kernel has associated. 21:39:58 Letiute: ok, gpart shows "geoms" with a partitioning scheme created on them, and partitions if those exist 21:40:05 rwp both do.. thanks. How to see sizes? 21:40:21 yuripv ok 21:40:23 Try: diskinfo -v /dev/ada0 21:40:25 if you didn't create partitioning scheme on a disk (or a geli device), it won't show in the output 21:40:40 ook 21:40:43 Or with your scsi disks: diskinfo -v /dev/da0 21:41:14 wonderful 21:41:18 Since the SATA adaptor driver creates adaX devices and the SCSI adaptor driver creates daX devices. 21:41:30 k 21:41:43 Also: geom disk list 21:42:36 BTW when I said "with your scsi disks" that was only because of the controller used. The same disks moved to the opposite controller would get the other naming. 21:42:46 oh 21:43:08 So a daX might be either SATA or SAS but pretty sure adaX can only be SATA. (All administrivial points.) 21:43:24 ok 21:43:50 Is mirroring (raid1) better or rsync better for least interval/near to realtime sync? latter scans the full dir tree (less efficient)? can I mirror drives to directories in a single drive (stupid but canI) ? send/rec is best. I am just asking for understandin 21:44:54 BTW... A simple disk performance benchmark can be had with: diskinfo -cit /dev/da0 21:45:08 ok... 21:45:41 but make sure to read the BUGS section in diskinfo(8) man page :) 21:45:54 hm.. 21:46:01 Also if the smartmontools port is installed then smartctl -i is very useful as well. And the utility of smartctl. 21:46:22 not -x ? 21:47:48 Letiute, I did not quite understand your question about rsync versus raid versus other. Though without knowing the question I would advocate ZFS due to all of the advantages there. 21:48:16 rwp I have a serious problem not going away... Samsung SSD 990 PRO 1TB percentage used : 9% in a week... 21:48:33 Firmware Version: 1B2QJXD7 21:48:41 So... Nine more weeks and then you are done! 21:49:03 "The steady state of disks is full." --Ken Thompson 21:49:46 rwp why is it happening like this? 21:50:22 Letiute, It's your system. I would ask you that question. What is consuming the space every week? 21:51:12 1000 browsers at a time working in /tmp maybe. That is the only reason. Could it be? 21:51:21 that I can think of 21:51:30 and a database (postgres) 21:51:41 but... nvmes.. /ssds don't wearup like that 21:51:51 Try: zfs list -o compression,compressratio,refquota,quota,used,usedds,usedsnap,refer,creation,name 21:52:30 And then look over the data. Should be able to tell which datasets/snapshots/clones are referencing data. 21:54:00 On a more local file system hierarchy level, "ncdu" is a favorite tool, use it to explore directory trees for disk usage. 21:54:07 rwp https://termbin.com/4cnc 21:55:05 Letiute, What did you want me to say about your dataset property list? I need a hint. 21:55:42 you asked for zfs list 21:56:00 I asked you to run "zfs list -o compression,compressratio,refquota,quota,used,usedds,usedsnap,refer,creation,name" and then for you to look at it. 21:56:40 That output you shared was not the above command. 21:57:46 Letiute: https://www.tomshardware.com/news/samsung-990-pro-firmware-update-released-ssd-health 21:57:48 Title: Samsung 990 Pro Firmware Update Addresses Failing SSD Health | Tom's Hardware 21:58:14 your firmware looks to be outdated 21:58:51 hm, no, looks up to date, was looking at 980 one (have that one) :D 21:58:57 yuripv, Good catch! 21:59:30 not so good, they seem to have the latest 22:00:29 thanks but the firmware update aint helping 22:00:32 "it seems as though 990 Pro owners who have already experienced significant health declines will need to RMA the SSDs to replace them" 22:00:42 what ist he best way to put tmps insetad of /tmp on disk? 22:00:59 yuripv, btw... The reason I like to do a quick benchmark is that by doing that I realized I had some disks on slow SATA ports on a particular motherboard. Only two of hte ports were 6Gbps and the other ports were 3Gbps and by realizing that I was able to address the inequality. 22:01:52 Letiute, I'm sorry but I do not understand your question. Please say more? 22:02:59 rwp I meant the disk is dying https://termbin.com/pd3g0 22:05:17 In only 241 hours of running? It should be under warranty then. What are the symptoms of it dying? Also run "smartctl -l error /dev/daX" on it. 22:06:18 In 10 days it has had 70TB of data written to it? That's quite a busy amount of data. 22:06:18 rwp Percentage Used: 9% in 2 weeks 22:07:54 I don't know your application but perhaps setting up and using a ram based file system would be useful? It all depends upon the details though. 22:09:00 I might also investigate extending the dirty cache sync time. Allow more time for cached data to remain unwritten. Maybe. I don't know. 22:09:02 if I run 1000 concurrent browsers (headless), the write in /tmp dir which is on drive. maybe thats how 70TB was written in 10 days? + a postgresql server with 2 queries concurrent. 22:09:23 For memory disks see section 19.10 of https://docs.freebsd.org/en/books/handbook/disks/ 22:09:24 Title: Chapter 19. Storage | FreeBSD Documentation Portal 22:11:16 Re: "1000 concurrent browsers" rather makes my head spin with the task of it. What is the need for such a monstrous farm of browser? 22:11:16 rwp ok but that does not tells how to use it as /tmp 22:13:24 I am sure if you search around you will find various helpful articles about it. I searched and quickly found some. Here is one. I haven't read it. https://tech.poojanblog.com/blog/unix-linux/putting-freebsd-tmp-and-var-on-a-memory-disk-md/ 22:13:26 Title: Poojan's Tech Blog / Putting freebsd /tmp and /var on a memory disk (md) 22:14:28 is there a service to activate tmpfs that would be used insetad of /tmp dir? 22:14:51 If it were me I would put those into a jail container and then use the memory disk /tmp in the jail. So as to keep those contained. 22:23:14 rwp can this commadn simply will turn /tmp from disk to ram? `tmpfs /tmp tmpfs rw,mode=777 0 ` ref: https://wiki.freebsd.org/TMPFS 22:23:15 Title: TMPFS - FreeBSD Wiki 22:33:12 `mount` says `zroot/tmp on /tmp (zfs, local, noatime, nosuid, nfsv4acls)` and `tmpfs on /tmp (tmpfs, local)`. Is that the verification that all the files in /tmp drive based are not in /tmp ram based? 22:33:57 what other places should be in tmpfs? /tmp /var ? any other? 22:37:37 use tmpfs for /var? 22:38:18 Is there any any difficulties / maintenance on mirrors ? I have a really hard time installing packages (download is very slow, all other network activity is as usual) 22:38:59 which mirror are you hitting? 22:39:48 I'm pretty new on freebsd so I can't tell but I guess 'the default mirror' 22:40:14 ServerName update.FreeBSD.org 22:40:37 is it about packages or freebsd-update? 22:40:41 right.. then it should say which 22:40:49 right 22:40:56 pkg.FreeBSD.org in /etc/pkg/FreeBSD.conf 22:41:18 do a `host pkg.freebsd.org` (or open in a browser), it should show the mirror name 22:41:20 about package, sorry for the confusion 22:41:49 pkgmir.geo.freebsd.org 22:42:11 and what's the IP? 22:42:18 or ptr? 22:42:25 ah, found it: pkg0.syd.FreeBSD.org 22:42:43 (australia) 22:43:09 hmmm... It may be my ISP.... 22:44:22 my 25 hop traceroute to their seems okay.. 22:44:26 er, zayo.com 22:46:22 there is a maintenance window today (in fact I work for my ISP and there is a maintenance today at our collect provider... should be it) 22:46:29 thanks 22:47:05 :) 22:47:12 cool 22:49:47 gman999 why use tmpfs in /var? is it like /tmp? 22:49:56 no.. you shouldnt 22:50:05 i was asking, not saying you should 22:50:08 i see 22:50:16 using tmpfs depends on context 22:50:23 what other directories to put in tmpfs? is the /tmp dir the only temporary/most written files dir? 22:50:36 why are you using tmpfs to begin with? 22:50:40 * Letiute wants to save his disk io 22:50:58 heavy read/writes. 22:52:46 so not an ssd? 22:52:55 i mean, wtih cf cards, etc i used a lot of tmpfs 22:53:00 in the past 22:53:24 /var/log can make sense 22:53:32 but i say "can"... not that you necesarily should 22:54:30 yes, /var/log 22:54:33 anyother? 22:57:56 that's prob more than enough 22:58:25 but it will impact write speed and again, anything in those mounts will disappear with reboots 22:58:35 and you have to make sure you mount then correctly 23:09:10 gman999 ok 23:09:26 how to see which files are taking most disk i/o? 23:09:39 iostat 23:09:43 ok 23:10:04 iostat(8) to be specific 23:10:08 have fun. 23:15:13 cannot trace exact files and data written by them 23:28:13 I would definitely *not* want /var/log to vanish on reboot. 23:28:32 ya. 23:28:44 but have to handle the huge io 23:31:18 I'd start by figuring out which files get a lot of IO, and what they're used for. The latter requires application-specific knowledge. 23:32:38 the answer is top -m io and fstat/lsof 23:32:41 ya but can't find a tool that tells me file paths 23:33:15 i just mentioned it to you less than ten minutes ago 23:34:10 `top -m io` is cool ;) 23:34:39 yes but didn't gave "file paths". I think fstat does 23:34:56 `truss /program` too 23:36:06 sysutils/dtrace-toolkit and truss from there is said to be better than truss 23:36:23