00:35:42 wheel group did not exist on HP-UX either. 00:42:31 I am primary a Linux user, I’ve been trying to setup a “service” using rc.d but I’m having issues. 00:42:50 I currently cannot figure out why my daemon won’t work, the issue is with `-u syncserver` 00:43:02 Command: `/usr/sbin/daemon -u syncserver -c -f /usr/local/bin/syncserver –config=/etc/syncserver.toml` 00:43:04 User created: `pw user add syncserver -d /usr/home/syncserver -s /bin/csh -m` 00:45:07 Is there a reason you're not using (name)_user as defined in rc.conf(5)? 00:45:55 Or are you using it? Because -u requires superuser privileges. 00:47:20 I'm mainly referenceing other rc.d scripts, and "Practical rc.d scripting in BSD" 00:47:41 didnt see (name)_user I'll need to look into that 00:48:05 mason, paul.webster@dev-11 ~ $ id root 00:48:05 uid=0(root) gid=0(root) groups=0(root),1(bin),2(daemon),3(sys),4(adm),6(disk),10(wheel),11(floppy),26(tape),27(video) 00:48:10 ah you are correct 00:48:18 I just assumed it was across the board 00:48:28 The article doesn't cover everything that's possible to do with rc.d scripting - that's laid out in rc.conf(5). 01:44:09 thanks debdrup I already had the user and group setup correctly, but didn't realise that was a setting rc.d would use itself 01:47:20 I finally got my service to start, but I have an issue that "needs" to be fixed. My service needs python with specific packages installed (via pip) how would I go about doing this? 01:51:41 For debugging (just to get the service running), I did a simple "pip install" as root I know this is "bad", 01:53:27 As a "user" I was using a `python3 -m venv venv` with a `export PATH="/usr/home/syncserver/venv/bin:$PATH"`, which worked great, but even though I exported the path it didn't work in rc.d 01:56:10 that prohibition will soon be enforced 01:56:32 (there is a PEP about it that we haven't implemented yet) 05:00:52 greetings 05:01:17 I have a question on FreeBSD kernel internals, more precisely signal subsystem 05:04:30 is someone here, please ? :-) 05:22:46 alexandru_goia: we are here; just ask 05:23:24 have a look on this 05:23:32 a Linux implementation 05:23:53 http://code.uriel.ro/inventica/angel.html 05:23:54 Title: Angel 05:24:02 my purpose is two-fold : 05:24:12 yes, angel() :-) 05:24:27 a) write a simple but non-trivial FreeBSD system call 05:24:43 b) re-implement on FreeBSD the Linux/Unix idea 05:25:29 i modified some code, in /usr/src/sys/kern/kern_sig.c , but it doesn't work 05:26:05 only if I send killsig(getpid(), SIGKILL); to my process - which calls first syscall(576); that is sys_angel() 05:26:07 it works 05:26:29 but if I send SIGKILL from external source, then it kills my process 05:29:34 do you have some expertise in FreeBSD sources ? more precisely signals ? 05:34:36 while i applaud your approach to stop using systemd, i'm not sure this is the solution i'd pick 05:41:09 rtprio: talking to me ? 05:43:17 yes. 05:43:38 curious about your changes to kern_sig.c but i don't know enough to tell you whats happening 05:45:40 int kern_sigaction() 05:46:17 int sigqueue_add() 05:46:29 i touched them briefly 05:46:41 basically removed extra checks for SIGKILL so thy're not uncatchable? 05:46:57 because there's a lot of them in that file 05:48:44 I want to find in the code where SIGKILL and SIGSTOP is taken into account (UNIX dogma says that they cannot be masked, caught, ignored, blocked) 05:48:58 as receiving signals 05:49:40 https://github.com/freebsd/freebsd-src/blob/main/sys/kern/kern_sig.c#L496 05:49:42 Title: freebsd-src/kern_sig.c at main · freebsd/freebsd-src · GitHub 05:49:42 I think kern_sigaction() is the function that *sends* the signals, including SIGKILL SIGSTOP 05:49:47 but i am not sure 05:52:52 maybe, working on kill(2) ? 05:53:02 and killpg(2) ? 05:53:17 or using curproc ? 05:53:27 with or without PROC_LOCK( ) 05:55:24 in Linux we have for the current process, the pointer struct task_struct *current; 05:55:47 in FreeBSD we have both struct thread; struct proc; and curproc and curthread 05:56:37 I don't quite understand the threads system in FreeBSD... 06:07:59 i donno man, hang around; maybe someone will answer when they wake up 06:12:24 hmm, they left 06:13:11 kern_* is conventionally the implementation of a syscall by the same name (it's actually, e.g., sys_sigaction(), but we commonly then call into kern_*() for details shared between different ABIs) 06:15:45 https://wiki.freebsd.org/AddingSyscalls has some good information in the area 06:15:47 Title: AddingSyscalls - FreeBSD Wiki 06:15:59 including a link to a presentation by Brooks 08:06:14 Hi 08:06:32 I just discovered that my snapshots for a really important dataset seems empty 08:07:48 https://0x0.st/HiXg.txt 08:08:02 why can't I check zfs snapshot content with .zfs ? 08:08:43 eoli3n: the .zfs directory can be disabled by default 08:08:51 but I didn't 08:09:02 it seems that my data are not written on the dataset 08:09:41 eoli3n: paste that `zfs get all zroot/encrypted/sync | grep snap` 08:10:07 eoli3n: do you have the encryption key loaded for this dataset? 08:10:21 https://0x0.st/HiX6.txt 08:10:21 yes 08:10:23 i do 08:10:26 the problem is not here 08:10:32 eoli3n: I'm not very familiar with that but that would be a good reason for not showing the content of datasets I guess 08:10:38 even before the snapshot problem it seems that i'm not writting on the dataset 08:10:45 that dir is mounted in a bastille jails 08:10:52 eoli3n: so yes I concur there's no data in the snaps 08:11:13 embiggen/usr/home/dch usedbysnapshots 87.3G - 08:11:13 embiggen/usr/home/dch snapshots_changed Wed Feb 22 10:49:01 2023 local 08:11:20 these 2 properties should be filled 08:11:24 https://0x0.st/HiXI.txt 08:11:57 before snapshot problem, i need to solve "where are my datas" :) 08:12:03 eoli3n: what happens if you manually create a snapshot? 08:12:12 as said 3 times 08:12:17 ^ 08:12:22 not a snapshot problem 08:12:27 no data in the dataset 08:12:48 zroot/encrypted/sync 200K 23.8G 200K /data/zfs/sync 08:12:51 empty dataset 08:12:56 but it is mounted 08:12:59 so this is nothing to do with zfs and everything to do with syncthing 08:13:00 i guess that bastille does something wrong 08:13:05 is there any data there at all? 08:13:12 no problem with syncthing 08:13:15 there are datas 08:13:22 ls -> camera/docs/screenshots/seedvault/sigal/ 08:13:27 missing spaces 08:13:34 camera/ docs/ screenshots/ seedvault/ sigal/ 08:13:43 ok so 08:14:14 same problem with my backups 08:14:17 this is a design problem 08:14:22 what I setup is not working 08:14:23 from the root, what do you see from `df -h /usr/local/bastille/jails/syncthing/root/data/zfs/sync` 08:14:42 this is to try and see what the upper filesystem is in that jail 08:15:10 datas are in that dir yes 08:16:02 and what does it show? 08:16:03 I sould ask on #bastillebsd 08:16:07 dch same dataq 08:16:11 same datas 08:16:16 ah 08:16:17 /jails/g01/var/cache/pkg 707G 845M 706G 0% 40k 1.5G 0% /jails/g03/var/cache/pkg 08:16:22 you meant df -h lets try 08:16:24 thats from a nullfs here 08:16:38 what we want to see is what does freebsd see as the upper volume 08:16:59 I'm wondering if bastille has (e.g.) started later, and mounted something on top of the nullfs mount 08:17:04 https://0x0.st/HiXG.txt 08:17:06 so writes don't go where you think 08:17:27 dch that's it yes, it write under the dataset 08:17:30 "under" 08:17:43 i need to redesign what i do 08:19:07 eoli3n: can you do specifically `df -h /usr/local/bastille/jails/syncthing/root/data/zfs/sync` ? and `zfs get all /usr/local/bastille/jails/syncthing/root/data/zfs/sync` 08:19:22 not the df -h | grep 08:19:40 the first will show precisely what the top layer filesystem is 08:19:55 toor@vps /d/z/sync# df -h /usr/local/bastille/jails/syncthing/root/data/zfs/sync 08:19:55 Filesystem Size Used Avail Capacity Mounted on 08:19:55 toor@vps /d/z/sync# zfs get all /usr/local/bastille/jails/syncthing/root/data/zfs/sync 08:19:58 oups sorry 08:20:01 https://0x0.st/HiXd.txt 08:20:10 the latter will tell if (assuming its the overlaid zroot/bastille/jails/syncthing/root "hiding" the original nullfs mount 08:20:17 np 08:20:44 hmm 08:20:50 thats not what I expected to see 08:21:24 FSTAB /data/zfs/sync data/zfs/sync nullfs rw 0 0 08:21:31 in my bastille template 08:22:07 the df shows me that its not a nullfs mount there, its a zfs one 08:22:16 and the zfs get all says its not zfs 08:22:17 huh 08:24:00 eoli3n: I think you should check this with bastille 08:24:22 yep 08:24:27 thanks for helping dch 08:24:44 np 08:29:32 I still haven't figured out my rc.d service issue; link service file https://bsd.to/Jh47 08:29:33 Title: dpaste/Jh47 (Bash) 08:29:35 good morning everyone im running freebsd 13.1 RELEASE -p6 amd64 on my lenovo x1 extreme gen 3 im trying to get suspend and resume working i can suspend with no problem but when trying to resume when opening up my laptop im not getting any video 08:29:43 I have a service I want to run “syncserver”, a rust binary app I compiled. 08:29:54 syncserver requires a python enviroment with specific packages available (installed through pip). 08:30:04 If I install py39-pip and `pip install` the packages then the service will run without issues 08:30:11 Therefor I assume that I’m incorrectly passing the python virtual environment to the rc.d service 08:30:34 qtc`: I don't think PATH makes it through daemon(8) 08:31:14 bsdbandit: are you using latest FreeBSD packages (i.e. not quarterly) ? 08:31:25 bsdbandit: if not, switch to those and see if it helps 08:32:11 bsdbandit: and then next stage is to see if you have any extra devices that are not resuming correctly. This is best asked in #freebsd-xorg channel imo 08:33:09 thanks 08:33:12 dch: 08:33:33 qtc`: I don't see any other ports doing a venv in their rc.d script, so I suggest you make a small shell script that sets up the venv and calls your stuff 08:34:47 updated https://bsd.to/oQxO 08:34:49 Title: dpaste/oQxO (Bash) 08:35:47 dch: do you need to create the python virtual enviroment on each boot? I though you need to only create it once. 08:36:09 qtc`: on each invocation, from memory. it sets up paths and some vars. 08:37:49 is that something that I would stick into `syncserver_precmd` ? 08:39:09 qtc`: you could try moving the venv activate into precmd, and see how you go 08:43:22 dch this is not a jail problem, i have the same behaviour with my backup dataset 08:43:26 which is not mounted in any jail 08:44:29 i think this is a key problem 08:44:34 because of encryption 08:48:29 but then... where are my data, if the jails are encrypted ? 08:48:33 datasets* 08:48:37 without the key 08:55:27 Or use the path in "bin/" or wherever inside the "venv" created instead of relying on "$PATH" 08:57:12 Latest failed attempt https://bsd.to/51Pc 08:57:14 Title: dpaste/51Pc (Bash) 08:57:45 how does a dataset handles writes when the encryption key is unavailable 08:58:06 parv: not sure what you mean 08:58:27 /path/to/venv/bin/kdlkldkfld 08:59:42 No need to create the "venv" again; use the path as "/tmp/venv/bin/python" 09:00:37 The failed attempt does list any error|warning messages or other issues 09:00:49 s/does list/does not list/ 09:01:20 Run that with "sh -xv ... >| log 2>&1" 09:01:29 adding /tmp/venv/bin/python still panics 09:02:37 What is "panic"? 09:05:14 eoli3n: no key loaded, the dataset shouldn't even be mounted IIRC. But ti definitely won't have data written to it (how would that be even possible) 09:05:30 I've only tried this casually, cribbed off https://klarasystems.com/articles/openzfs-native-encryption/ and it just worked (tm) 09:05:31 Title: OpenZFS Native Encryption | Klara Inc 09:05:33 dch but i does write 09:05:38 talking on #zfs if you want 09:06:20 I might follow along there, I'm not really clear what the issue is anymore. nullfs is irrelevant, question is "my dataset is mounted, there is data in it, but snapshots are empty"?" 09:06:29 anyway I have an errand, 09:07:16 sh -xv /usr/local/etc/rc.d/syncserver || 3623 Lines of output 09:07:25 That encrypted dataset may not be mounted but if something else would mount along that same mount point, the data would go in a plain directory somewhere up the hierarchy of the dataset 09:07:53 parv: syncserver is a rust binary file, it panics if it doesnt have its python 09:09:40 qtc`, Do you mean "syncserver" is looking for a Python interpreter accessible via literal "python" without an absolute path? 09:11:16 probably, I don't know for sure but thats my impression 09:12:43 if I `pip install ` via py39-pip, then everything works as expected, but I know thats bad so i'm looking to doing it correctly 09:12:52 qtc`, Could you post (the output from "sh -xv") some 10 lines before & after when "activate" is sourced & when "/usr/local/bin/syncserver" is actually run? 09:13:46 i'm a python newbie but shouldn't you use pip --user of a venv? 09:14:08 or a venv 09:20:15 qtc`: doesn't syncthing already have an rc script? 09:21:49 qtc`, I take it the modules needed for "syncserver" are not available as FreeBSD packages|ports? 09:25:34 rtprio, Looks like qtc` is working with https://github.com/mozilla-services/syncstorage-rs 09:25:35 Title: GitHub - mozilla-services/syncstorage-rs: Sync Storage server in Rust 09:25:47 parv: no they are not 09:26:03 qtc`, Aye. Thank 09:26:13 s 09:32:00 `sh -xv` output https://bsd.to/QRuJ 09:32:01 Title: dpaste/QRuJ (Bash) 09:32:33 parv: the first block is probably useless 09:33:52 you'll notice that the service started and I needed to kill it, thats bc I ran `sh -xv /usr/local/etc/rc.d/syncserver` form the terminal as a user (which works) 09:35:41 Did you start in a spanking new environment/terminal where your existing environment would not any effect? 09:36:44 new tab > ssh > truenas > jail console > run command (root) 09:36:58 same will happen if I run as syncserver user 09:43:56 qtc`, After line 58 at https://bsd.to/51Pc, add "env; exit" to see if "activate" has sufficiently stuffed the environment or not 09:47:17 qtc`, Sorry make that "env | grep -E 'PATH|PYTHON' ; exit' 09:49:57 I can already tell you there is no python https://bsd.to/qaEv 09:49:58 Title: dpaste/qaEv (Bash) 09:50:21 oh 09:51:03 I copied the raw of https://bsd.to/51Pc, and edited it as suggested 09:51:07 qtc`, Would that not be found under /tmp/venv/bin ? 09:51:41 yes python is /tmp/venv/bin/python 09:51:58 So "python" is in PATH. 09:52:50 Does "syncserver" not have a debug|trace|verbose option that could tell why it would fail? 09:53:15 (Or use "truss", "ktrace") 09:55:37 There might be some Rust tooling (am not familiar with the ecosystem yet) 09:56:05 command_args="--config=${syncserver_config}; python -m pip install --upgrade pip" 09:56:07 sh: python: not found 09:56:58 but if i run pip where `env; exit` is inserted, pip runs as expected 09:57:09 hi all.. i was getting bug while i was compiling the nix-2.3.10. this is the log: https://pastebin.com/EreZr6hb 09:57:10 Title: ld: error: undefined symbol: Aws::Crt::Endpoints::RuleEngine::~RuleEngine()>>> - Pastebin.com 09:57:34 how to fix it? 09:58:12 qtc`, When you wrote " i run pip where `env; exit` is inserted, pip runs as expected", do you mean the "pip" installed system wide? 09:59:17 checking one sec 09:59:45 Alif, Have you installed all the required dependencies? 10:00:08 parv: yes, why? 10:00:54 Alif, Well to be sure some library/header were not missing 10:01:00 parv: this is my log https://pastebin.com/tP6bgtpa 10:01:01 Title: [root@klandestin /usr/ports/sysutils/nix]# MAKE_JOBS_UNSAFE=yes make===> Lice - Pastebin.com 10:01:37 Alif, Ok 10:02:05 parv, would you try compile it? 10:02:27 `which pip` is giving me `/tmp/venv/bin/pip` same for python 10:03:14 rust panic https://bsd.to/FUmh, due to missing python 10:04:30 looks like where the command_args are concerned python or pip dont exist 10:06:40 qtc`, Perhaps need to investigate if "run_rc_command" purges the environment 10:09:42 Alif, the current version in ports is 2.3.11, not 2.3.10. Try updating your ports tree and see whether the error goes away. 10:10:21 Alif, sysutils/nix was updated to 2.3.11 on 202208; from https://pastebin.com/tP6bgtpa seems you are building 2.3.10 10:10:23 Title: [root@klandestin /usr/ports/sysutils/nix]# MAKE_JOBS_UNSAFE=yes make===> Lice - Pastebin.com 10:10:31 mine packages are up to date ( i use freebsd 13.1-freebsd 10:11:01 parv: thank you. i will try 10:11:26 Alif, Not package, but the ports tree that would need to be updated as V_PauAmma_V wrote 10:15:59 parv : i did "portsnap fetch" but i got same version. https://pastebin.com/LgbfGNs5 10:16:00 Title: root@klandestin:/home/alif # portsnap fetchLooking up portsnap.FreeBSD.org mir - Pastebin.com 10:18:01 qtc`, Try shoving the environment variables needed in "syncserver_env" variable; refer to "/etc/rc.subr" for any details 10:28:36 parv: how to fetch updated port? 10:30:48 Alif, I personally use "git"; you may want to do a "shallow clone". BTW I just updated the ports tree and the latest commit at the time was "2023-0308 -0800 52e94aebf8 HEAD -> main, origin/main, origin/HEAD science/phonopy: Update 2.16.3 ..." 10:31:17 ... "52e94aebf8" is the commit hash 10:32:05 chmod 777 -R /tmp/venv/ -- syncserver_env="/tmp/venv/bin" -- env: /tmp/venv/bin: Permission denied 10:33:48 Use the whole of the "PATH=..." string 10:34:40 Also, please do not set "o+w" 10:36:46 Alif, shallow clone of the ports tree: "git clone --depth 1 https://git.FreeBSD.org/ports.git /somewhere/local/filesystem/ports" 10:36:47 Title: ports - FreeBSD ports tree 10:37:57 I got it working https://bsd.to/0JLy! 10:37:59 See also https://docs.freebsd.org/en/books/handbook/ports/#ports-using 10:38:00 Title: Chapter 4. Installing Applications: Packages and Ports | FreeBSD Documentation Portal 10:38:04 Thank you for your help parv 10:38:25 qtc`, Sweet 10:38:29 I'll destroy my jail and rebuild it correctly this time 10:39:41 parv: forget me for being dump. i have not portsnap extract.. ok i got newer version, but i cant compile because the newest commit does not bring nix source code 10:39:45 so i can't compile it 10:40:32 parv: this is my log https://bsd.to/108S 10:40:33 Title: dpaste/108S (Plain Text) 10:41:19 I though this would only take an hour, it took a day instead. 10:41:50 lol 10:43:10 hello again 10:43:15 what i missed? 10:43:21 Alif, Either mark /usr/ports/distfiles, & possibly other directories, to be writable as your user or run the command as root 10:43:53 ok.. i will run portsnap extract as root 10:46:33 parv: i can't find nix source code on https://cgit.freebsd.org/ports/tree/sysutils/nix 10:46:35 Title: nix « sysutils - ports - FreeBSD ports tree 10:46:41 what i should do now? 10:47:32 What does "cd /usr/ports/systutils/nix && make fetch" do as root? 10:48:28 *fudge* corrected: cd /usr/ports/sysutils/nix && make fetch 10:49:21 thank you.. it works on me 10:50:49 Alif, Aye. After "make fetch" finishes, use "make install" 11:02:43 i still got same error about undefined symbol aws:crt (( log: https://pastebin.com/s6dzJsJ3)) 11:03:20 i downloaded all build, test, library dependencies via pkg (https://www.freshports.org/sysutils/nix/) 12:57:47 Is there anything wrong with creating an new group for allowing users to mount removable media instead of using the "operator" group as described in the Handbook? 13:17:12 hello antranigv. I read the readme of jailer project, sound really good. Is there a way to switch from iocage managed jails into jailer? For now, I use FreeNS on myy NAS, with iocage. 13:23:41 bsdbandit: when I installed my thinkpas X280, I read and use this blog post: http://hauweele.net/~gawen/blog/?tag=suspend 13:23:44 Title: suspend | Gawen's blog 13:32:01 msiism: nope, go for it 13:32:16 Thanks! 13:32:27 And, would it be unwise to name that group, say, "usermount"? 13:32:40 Lovis_IX: in general this should be easy enough (tm) but FreeNAS will likely complicate matters significantly. 13:32:55 I'd start off with a small jailer setup on a separate box (laptop or something easy) 13:33:29 iocage has 2 main things 13:33:32 - a config file with all the jail-related settings in it which you'd need to transfer over manually between tools 13:34:06 - a root volume with all the files in it. you can probably just tar that up and untar it over the top of the jailer created config, then go through and tidy up the settings 13:41:32 otis, hi! Thanks for zabbix 6.4! Look like it support PHP 8.2: https://www.zabbix.com/rn/rn6.4.0 13:41:33 Title: Release Notes for Zabbix 6.4.0 13:41:39 https://support.zabbix.com/browse/ZBXNEXT-8204 13:41:41 Title: [ZBXNEXT-8204] PHP 8.2: Dynamic Properties are deprecated - ZABBIX SUPPORT 13:42:07 But I didn't tested this. 13:43:42 hello dch, thanks for your advises, I will try soon as possible (for now I have a ccat on my knee, it's difficult to concentrate) :-) 13:43:54 Lovis_IX: I envy you! 13:50:04 for the jail migration or for the cat? :-) 13:52:06 Lovis_IX hey there! not yet, with 0.1.3 (the next-next release) we will have a parser to migrate things from IOCage as well as BastilleBSD. 13:52:50 thanks antranigv 13:53:26 Lovis_IX but dch is right, things can be done manually as well. I haven't used it in a while, so it will take me couple of weeks to write a converter :D 13:56:27 Lovis_IX thank you for testing it! 14:00:12 I'm waiting for new disque with higher capacity, I will cange freeNAS for FreeBSD 13.1 then try to migrate jails from iocage to jailer. 14:00:41 change* 14:07:40 VVD: yep, it does. i forgot to remove that restriction after my testing. 15:14:09 Is there a way to determine the file system type on an external storage device before mounting it? 15:14:38 file(1) 15:18:36 Okay, thanks, `file -s ` does what I want here. 15:29:22 Now, file(1) says, "reading special files may have peculiar consequences". What do I need to watch out for? 15:30:43 whether your tape is rewound on closing, after this read 15:31:31 I'm not using tape yet. 15:32:09 The question is mainly about external HDDs and flash devices. 15:35:48 msiism, it's probably about things like terminals, which block waiting for input and consume it once it's available 15:35:57 it's a question for the device-driver, more than for the hardware device. Traditionally, actions like that were coded by having multiple special-device names in the filesystem for a single hardware device 15:47:13 msiism: also fstyp(8) might help 15:49:20 I see. Thanks. 15:50:52 fstyp for the win. :) 16:11:50 not neccesarily. doesn't identify apfs along with plenty more fs's 16:14:09 Okay, noted. 16:17:49 vdi, qcow(1,2), apfs, ... 16:18:06 it could use some attention 16:18:30 For USB keys, I almost always use FAT, though. 16:18:36 but its a good tool none the less 16:20:33 found myself plenty of times just using hd(1) along with hexedit to inspect the first bytes of a filesystem just to determine the proper tool to determine what it was 16:21:28 only to come out to the fact that google and such will give you a hint and lead you down the right path 18:27:14 antranigv: I wasn't able to join the jails chat this morning due to a meeting conflict :>( 18:27:53 rtyler not too late to join! shall I DM the link? 18:30:06 can anyone join? 18:30:40 antranigv: unfortunately not, still meeting, just multi-tasking ;) 18:34:23 domlaut: you're curious too :-) 18:39:16 Lovis_IX: I'd gladly listen in, esp if it touches on jailer 18:39:16 domlaut yes 18:39:24 domlaut shall I DM? 18:39:39 yep 19:03:55 ugh, ok so i copied over my zfs filesystems from one to the other, but i can't seem to boot. 19:05:21 Demosthenex: did you make the new disk bootable? man zfsboot or gptzfsboot 19:06:45 so i made a fast base install to the new laptop, booted it OK, then booted USB and removed all the filesystems in zroot. then i zfs sent all the ones from my other box, and confirmed they were present 19:06:50 rebooting didn't find the boot i guess 19:07:54 and the zfs tree looked the same as the fresh install? 19:07:55 which is odd, because i left the efi partition alone from the new install 19:08:14 i wonder if the old system zpool name of zssd vs the new install of zroot is an issue 19:08:27 yes, that could very well be 19:09:32 though after the geli initialization, it's not loading the boot screen. 19:09:39 makes me wonder about the loader 19:11:13 hrm, this lsdev shows my zroot 19:11:40 what's the error when the laptop boots 19:12:28 can't find efi something, reboots in 4 sec, i hit a key now at a prompt. 19:12:32 i can lsdev, lszfs, etc. 19:14:27 hand repairing the zfs loader is something i haven't done for quite a while 19:14:51 oh, i wonder if the old lappy was BIOS boot, not uefi 19:15:07 meaning my kernel is in zroot/ROOT/boot/... 19:15:10 Demosthenex: did you set bootfs on root pool? 19:16:15 (if it were me, i'd do the fresh install, import the zfs, but manually integrate them into the fresh system 19:16:47 yuripv: not sure. must check. 19:16:49 but importing with zfs-recv the zfs partitions should be under the existing pool, right 19:17:04 existing pool is uefi 19:17:07 old pool may be bios 19:17:27 "show" has all the right things for the boot env, etc 19:18:51 i mean the "bootfs" property on your boot pool, e.g.: https://pastebin.com/cDZKxB2y 19:18:52 Title: $ zpool get bootfs systemNAME PROPERTY VALUE SOURCEsystem - Pastebin.com 19:19:47 yes, i have to boot to usb to find out, just a moment 19:20:09 that is set on the old 19:21:36 no, new has no bootfs. 19:21:37 fixing 19:22:22 but the installer would have set it bootable 19:23:25 wonder if when i removed the old os filesystems it cleared that 19:24:22 ah, now i got the loader. 19:26:19 and i booted, correcting fstab. 19:28:09 ha! love it! it all came right up 19:28:13 stumpwm in all it's glory 19:29:31 ok, so how should i update the bootloader just in case 19:34:20 cp /boot/loader.efi /boot/efi/efi/freebsd/loader.efi 19:34:45 (if you have the now-default mount of esp) 19:35:42 https://forums.freebsd.org/threads/update-of-the-bootcodes-for-a-gpt-scheme.80163/ is a good guide 19:35:44 Title: Update of the bootcodes for a GPT scheme | The FreeBSD Forums 19:35:49 that was actually pretty smooth 19:36:06 so tomorrow i can shutdown to single user, take a final snapshot, and incremental update by zfs send, and i'm done 19:36:09 sweet 20:09:17 Hrm. Anyone know what creates the /zroot directory on a ZFS-on-root FBSD system? 20:09:41 Apparently, all my FBSD systems have a /zroot directory except one and I can't figure out why. 20:11:43 the installer does, because that's the name of the pool 20:11:55 It's the mountpoint for the zroot dataset. 20:12:14 It's used to make boot environments work, I believe. 20:13:59 Strange. This was setup as ZFS on root from day one. Everything works as expected but there isn't a /zroot directory so automated snapshots fail testing for /zroot/.zfs/snapshot directories. 20:14:09 I'm stumped as to how that could not exist. 20:14:43 It's not part of zpool-history(8)? 20:15:03 Rather, it's excluded from that? 20:15:48 There should be a `zfs set mountpoint=/zroot zroot` 20:18:15