00:12:19 <dgriffi> anyone here run freebsd at linode?
00:13:43 <Erhard> BSDode
00:15:19 <markmcb> what is the goto VPS provider for freebsd? vultr?
00:15:54 <Erhard> Yuo get a three free forever ones from Oracle Cloud. But they aren't high spec
00:17:42 <Erhard> Kamatera and Contabo also run FreeBSD well.
00:18:04 <Erhard> But I am sure there are hundreds.
00:25:16 <meena> dgriffi: if it doesn't run, tell me, and I'll try to get it fixed in cloud-init
00:26:47 <dgriffi> I've been fiddling with freebsd on linode for a few years.  lately I've been doing more on that machine with gitlab and I realized that the extreme slowdowns are because connections try to use ipv6, fail, then fall back to ipv4.
00:27:08 <dgriffi> so I fiddled with it some more and managed to complely break networking.
00:27:23 <dgriffi> so I reinstalled from scratch and I'm back to where I started.
00:28:31 <dgriffi> as far as I can tell, ipv6 is working.  my machine just can't talk to anything using ipv6.
00:30:23 <meena> that's bizarre
00:32:12 <Erhard> What happnes when you mtr -6 to, say, google?
00:32:45 <meena> does linode / akamai have a free tier?
00:33:25 <dgriffi> "mtr"?
00:33:31 <Erhard> There are a lot of free offers for a limtied time and free $100 worth on linode
00:33:50 <dgriffi> morte_: not to my knowledge.  I pay for my account.
00:33:52 <Erhard> It's a fancy traceroute, after using it you will never go back
00:34:19 <Erhard> You can use traceroute6
00:35:08 <Erhard> (the nox variety of mtr... uses ncurses)
00:35:58 <meena> dgriffi: it's "meena".
00:37:04 <dgriffi> traceroute6 to google.com (2607:f8b0:4005:813::200e) from 2600:3c01...
00:37:18 <dgriffi> so dns is giving me v6 addresses...
00:37:34 <dgriffi> but I'm getting three asterisks for each hop
00:37:46 <Erhard> I think that can be resolved using ipv4, though
00:38:11 <Erhard> I would check with linode and see what ip and gateway you are supposed to use for ipv6
00:38:22 <dgriffi> doing a ping6 of my ipv6 address works fine.
00:38:23 <Erhard> And if you should be using a static ip or dhcp6
00:38:38 <Erhard> From where? Locally ?
00:39:06 <Erhard> Do you have a default route for ipv6 ?
00:39:11 <dgriffi> over in #linode they told me to use fe80::1 for the gateway.
00:39:47 <dgriffi> Erhard: not sure.  I couldn't figure out where to plug in the ipv6 gateway
00:40:03 <Erhard> I defer to someone else more expert in ipv6. I have oly set up a few systems with it.
00:40:29 <Erhard> There should be some tutorials online though, or in the handbook.
00:40:53 <dgriffi> Erhard: I've been combing throught them to no avail
00:41:09 <Erhard> Somehow you ened to get a default route in the routing table, and you need to have an ip configured. And then linode needs to be setup on there end to handle these... just like ipv4, just different ips.
00:41:25 <dgriffi> I found a bunch of them where I get down to the bottom and someone's posted "that doesn't work anymore for 13.x"
00:41:27 <Erhard> Do you use dhcp
00:41:30 <Erhard> ?
00:41:37 <dgriffi> yes
00:41:47 <Erhard> And you have that configured for ipv6 ?
00:42:04 <Erhard> I think that is a separate setting, IIRC.
00:42:12 <Erhard> Again, I am no expert on ipv6
00:42:31 <dgriffi> not sure
00:42:48 <dgriffi> the tutorials I've found are all really vague
00:54:50 <Erhard> On a ipv6 vps I have this is how it is setup in rc.conf
00:54:52 <Erhard> https://bsd.to/UBLq
00:54:54 <VimDiesel> Title: dpaste/UBLq (Plain Text)
00:55:09 <Erhard> USe your ip of course.
00:56:22 <Erhard> I found a slightly older post saying you needed the dhcp client from ports to do ipv6, but I bet that is no longer the case. I do not know, though.
01:15:43 <Kangie> Hi All. I'm hoping that someone can help me understand why 'ifconfig mlxen0 -vlanhwtso' doesn't seem to have any impact on my ConnectX-3 card (mlx4en driver)? FWIW I'm on opnsense; so I'll happily wait for someone in that community respond if that's more appropriate.
03:16:16 <skered> kevans: Oh ok so _TIMEOUT isn't in the picture at all now.. good.
03:19:14 <kevans> skered: right, now it's more closely aligned with how libfetch timeout worked
03:19:32 <kevans> with the somewhat subtle difference that it's now possible for a transfer to be so slow that we consider it for timeout still
06:21:47 <concrete_houses> so why did chrome get booted?
06:21:57 <concrete_houses> something about ad blocker
06:22:41 <ScottTheGeek> People use Chrome? :P
06:23:20 <TommyC> concrete_houses: google said they were going to remove ad blocker addons or something like that (essentially they were going to disable adblockers somehow)
06:23:51 <TommyC> people don't want to get free viruses and find out about single moms near them so everyone jumped ship from chrome
06:24:07 <TommyC> seems the Windows people are somewhat divided amongst Opera GX and firefox atm
06:26:27 <ScottTheGeek> Opera is just a re-skinned Chrome as far as I'm concerned
06:28:53 <TommyC> ScottTheGeek: I think that's regular opera. Opera GX is their new "gaming" browser or whatever
06:29:19 <RhodiumToad> booted?
06:29:48 <ScottTheGeek> Does GX actually use a different rendering engine/core?
06:30:17 <AlaskanEmily> I think it's purely some shell adjustments.
06:30:34 <AlaskanEmily> Also I think it includes some cryptowallet by default stuff too?
06:31:03 <TommyC> ScottTheGeek: I've no idea, I've used firefox since elementary school (that was like firefox 3.0 or something)
06:31:03 <ScottTheGeek> why does everything have to be "crypto" or "blockchain". It's getting tired.
06:31:39 <ScottTheGeek> I've used Netscape -> Phoenix -> Firebird -> Firefox
06:31:56 <TommyC> oh yeah, I forgot about netscape
06:32:08 <TommyC> but that was back when I was just learning to type on Windows 95
06:32:11 <ScottTheGeek> the DNA is still there in Firefox
06:32:35 <ScottTheGeek> Heh, I just created a WFW3.11 vm for shits and giggles today
06:34:44 <ScottTheGeek> I learned to type on a Commodore PET
06:39:25 <grahamperrin> egl: a nit: "pools" (not "zpools"). It's a common misnomer.
07:30:48 <concrete_houses> lol
07:51:31 <bapt> RhodiumToad: ping
07:51:40 <RhodiumToad> ?
07:51:54 <bapt> do you have the ability to test a patch?
07:52:03 <bapt> for the CPU and memory extra consumption of pkg?
07:52:19 <RhodiumToad> sure
07:52:57 <bapt> https://people.freebsd.org/~bapt/patch-faster
07:53:04 <bapt> put that in your ports-mgmt/pkg/files
07:53:11 <bapt> and tell me if it helps in your case
07:58:05 <RhodiumToad> ok, will do. need a few minutes
07:58:21 <bapt> thanks
08:03:24 <RhodiumToad> ugh, "process with pid 783 still holds the lock" - is that from pkg or sqlite?
08:04:18 <RhodiumToad> I'm guessing sqlite
08:05:54 <RhodiumToad> dammit, anyone know offhand how to force sqlite to break that lock? pid 783 is an unrelated process
08:07:11 <RhodiumToad> (the old problem of a low PID being allocated to a long-running daemon after reboot, confusing anything that uses pids as part of a locking protocol)
08:11:37 <parv> If PID is being read from a file, remove the file?
08:12:41 <RhodiumToad> it seems not to be a separate file
08:13:26 <bapt> this is pkg telling you that ?
08:13:46 <RhodiumToad> yeah
08:14:39 <bapt> pkg shell SELECT pid FROM pkg_lock_pid;
08:14:55 <bapt> pkg shell "SELECT pid FROM pkg_lock_pid;"
08:16:00 <RhodiumToad> hm, different pid now, but that's because I just tried pkg add again and got an OOM kill on it
08:16:05 <RhodiumToad> this is pre-patch
08:16:13 <bapt> ok
08:16:18 <RhodiumToad> just getting a baseline before testing the patched one
08:16:23 <bapt> sure
08:16:39 <bapt> maybe make a baseline based on 1.19* :D
08:19:17 <RhodiumToad> I don't think I have a package of that handy on this vm
08:21:47 <RhodiumToad> ok, the package that was OOMing on 1.20.1 can now be added fine, with ~85MB max rss used according to time -l
08:24:08 <RhodiumToad> tried a more complex package with more dependencies and the max rss was ~90MB
08:25:31 <RhodiumToad> ok, at a convenient moment I'll restart the poudriere build I have going in another VM and try it with this patch, see if it helps the speed of _depends phases
08:31:58 <bapt> from my testing pkg 1.20.2 + my patch is as fast as pkg 1.19.*
08:53:10 <parv> RhodiumToad, bapt, Are you testing anything specific in "pkg"?
08:54:35 <RhodiumToad> cpu and memory usage in pkg add
08:54:47 <RhodiumToad> (as used heavily by poudriere, and ports builds)
08:57:02 <parv> Thank you
08:58:40 <martinrame> Hi, any smb4 expert out there?.
08:59:36 <martinrame> I need to create a passwordless share and I've tried everything, but the server allways asks for a password.
09:00:55 <mage> you need to setup Samba for this
09:01:34 <mage> SMB* stuff in BASE are totally outdated
09:01:52 <martinrame> mage: yes of course. I have samba running on this server for more than a year.
09:02:11 <martinrame> But now I need to setup a passwordless share
09:19:03 <mage> isn't setting public = yes sufficient?
09:20:26 <martinrame> mage: apparently no.
09:21:34 <martinrame> mmm, I think I found the culpit!: "reject guest session as encryption is required
09:22:44 <martinrame> yes!
09:48:39 <grahamperrin>  > pkg: No SRV record found for the repo 'FreeBSD'
09:48:39 <grahamperrin> <https://bsd.to/mRjA/raw>
09:48:41 <VimDiesel> Title: mRjA
09:49:49 <grahamperrin> I assume that the message is informational (not a warning). True? ▲ bapt
09:50:09 <bapt> this is a warning
09:50:35 <bapt> pkg -vv
09:50:45 <bapt> I need the full description of the repositories
09:50:50 <bapt> right
09:51:01 <bapt> you change the url of the FreeBSD repo
09:51:08 <bapt> but you have kept the mirror type to SRV
09:51:16 <bapt> so it cannot work
09:52:14 <bapt> I decided to make pkg more strict in that area starting 1.20, because of too many reports of people messing up like you right now
10:30:07 <duncan_bayne> Hey :)  Just trying a fresh install of 13.2 and the virtualbox-qt package seems to be gone.  Is there an obvious way of viewing package history that I'm missing?  Looking to see the history of why / how it got removed (with an eye to figuring out when it might be back ;) )
10:39:18 <duncan_bayne> Can't find any relevant history in the Github ports mirror
10:41:17 <mage> ZFS question: have there been any improvements regarding concurrent snapshot/delete/bookmark/cursor between FreeBSD 12.x and FreeBSD 13.x?
10:42:02 <mage> I'm asking this because for some unknown reason zrepl is super slow with ZFS 12.x (STEPPING takes forever)
10:42:06 * grahamperrin can not join #pkg
10:47:12 <Demosthenex> hrm. so i'm setting up ejabberd, and i got an acme cert, but on startup ejabberd is complaining it doesn't recognize the CA which issued the cert. so that means ejabberd isn't understanding /etc/ssl/certs. i added the ca_root_nss package, and tried pointing the ca_file to /etc/ssl/cert.pem, and stil no dice.
10:47:40 <Demosthenex> i did confirm the let's encrypt cert is in /etc/ssl/certs, and have run certctl rehash, still doesn't recognize
11:00:50 <meena> Demosthenex: that new?
11:03:06 <Demosthenex> well ejabberd has built in acme now, so i don't need other tools
11:03:15 <Demosthenex> the cert works. thats ok
11:03:23 <Demosthenex> its just ejabberd logging a message that it doesn't recognize the root ca
11:03:49 <Demosthenex> and despite adding the ca_root_nss package, and tryign to use ca_file /etc/ssl/cert.pem, it still doesn't recognize it
11:05:03 <meena> Demosthenex: so… if it works, then what CA store is it using?
11:05:18 <meena> is ejabberd linked against OpenSSL?
11:16:45 <Demosthenex> i'm honestly not sure. it doesn't tell me :P
11:45:54 <bapt> RhodiumToad: tested on the freebsd cluster, 3 times faster
11:46:02 <bapt> that is imho good enough for a 1.20.3
12:57:09 <ridcully_> Demosthenex: shot in the dark: might be using what erlang is giving it
13:00:01 <RhodiumToad> bapt: it's looking better inside poudriere, too
13:00:41 <bapt> this is poudriere in the cluster
13:01:10 <RhodiumToad> well, I meant in my own tests
13:04:42 <Demosthenex> ridcully_: not sure what that would be. i tried a few times, and since it doesn't seem to prevent it from working, maybe it doesn't matter ;]
13:05:06 <RhodiumToad> bapt: regarding provides/requires in principle, did you see what I said yesterday about having two packages with the same provides?
13:07:27 <bapt> RhodiumToad: yes but it they are incompatible then we as a project should not allow both to have the same soname
13:12:33 <RhodiumToad> think  /usr/local/lib/guile/2.2/extensions/libfoo.so  and  /usr/local/lib/guile/3.0/extensions/libfoo.so
13:13:23 <RhodiumToad> relying on sonames that are unique independently of paths is what got Windows a bad name, let's not make the same mistake
13:13:25 <bapt> then libfoo.so should never ever be in the shlibs_provide
13:13:40 <RhodiumToad> right, but I didn't put it there, something in the port build did
13:13:49 <bapt> there is a knob for that
13:15:23 <RhodiumToad> which is?
13:15:24 <bapt> BUNDLE_LIBS=yes
13:16:16 <bapt> we should have something more fine grain
13:16:23 <RhodiumToad> from its description, that's nothing like what's appropriate here
13:16:32 <bapt> it is
13:16:35 <RhodiumToad> this isn't a bundled lib, this is exactly what the port is building
13:16:47 <bapt> is prevent the libs to be put in shlibs_provided
13:17:04 <bapt> btw why /usr/local/lib/guile/2.2/extensions/libfoo.so has a soname?
13:17:45 <RhodiumToad> because the upstream's build procedure did it?
13:18:09 <RhodiumToad> but in fact it needs one, because of its clients, one of them needs to link it directly
13:18:21 <RhodiumToad> (while the other one dlopen's it)
13:18:36 <bapt> you have things directly linked to it?
13:19:35 <RhodiumToad> yes (using rpath, so not needing it to be anywhere that ldconfig has to deal with)
13:20:20 <bapt> yes shlibs_provides should only be for packages with USES_LDCONFIG
13:21:00 <RhodiumToad> ah. in this case I can probably turn that off
13:22:59 <RhodiumToad> ok, that's useful, because it had USE_LDCONFIG from when a version installed into /usr/local/lib directly
13:23:18 <RhodiumToad> had to move that to avoid a conflict
13:24:22 <RhodiumToad> that doesn't explain how it ended up with a shlibs_provide that wasn't in the ldconfig default directory, though?
13:24:53 <RhodiumToad> shouldn't it be looking only for shared libs in such dirs, and assuming ones elsewhere are not suitable for _provide?
13:34:56 <debdrup> Freaky: Jessica (using her as an example, although there are others) has a bunch of commits mentioning GENERIC_LIBCOMBAT which include "commit <hash>" in the log description; is it possible that something can match on that and make them into links?
13:36:15 <kenrap> (She did a pretty good job finding and cleaning out that COMPAT stuff in those commits imo)
14:14:40 <Freaky> debdrup: I've made it match on "commit <hash>" and on anything that looks sufficiently like a hash (a-f and at least one number)
14:53:07 <VVD> Is it possible to "Maintainer timeout" this PR: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=271499 ?
14:53:10 <VimDiesel> Title: 271499 – porters-handbook/versions: missing entry for LLVM 13 for stable/13
15:18:14 <debdrup> Freaky: thanks!
15:35:21 <hwpplayer1> hi FreeBSD people !
15:37:23 <RhodiumToad> hello
16:01:22 <skered> kevans: Just need that in a release now... My vbox builder was able to get thru the 30s gauntlet so I'm good for now.
16:36:04 <kevans> skered: it's in 1.20.3 that just dropped within the last 12 hours
17:14:55 <concrete_houses> why is node so popular?
17:15:01 <concrete_houses> does node work on freebsd?
17:15:12 <concrete_houses> java script seems hell
17:15:35 <concrete_houses> too many MBA looking a gui as fancy toy? not thing to indicate what customer wants?
17:15:56 <RhodiumToad> node is available as a package or port
17:16:46 * RhodiumToad wouldn't use it, but it gets pulled in as a dependency by stuff he does use
17:20:17 <concrete_houses> any adivce to get job now in boston toad?
17:20:21 <concrete_houses> this market seems tight
17:20:29 <concrete_houses> are we going into recession deeper?
18:01:50 <markmcb> anyone have insights on why rpi-firmware lags so much (about 2 years from upstream)? it's not clear if it's just "no one's done it" or if there's a blocker.
18:06:07 <markmcb> or maybe a better question, where should i look to try to answer that myself? mailing lists?
18:10:46 <markmcb> I think I found the answer: https://lists.freebsd.org/pipermail/freebsd-arm/2021-March/023214.html
18:10:48 <VimDiesel> Title: RPi4 Status and sysutils/rpi-firmware
18:11:44 <concrete_houses> what pkg lets libreoffice do spellcheck?
18:12:26 <concrete_houses> are 1 page resumes with financial rsult better?
18:12:27 <kevans> markmcb: I mean, we kinda pull changes as something pertinent looking happens (as it's brought to attention)
18:12:32 <concrete_houses> liek savce 4mil or gaine 5mil
18:13:01 <kevans> although with -current and maybe 13.2 we could maybe start tracking upstream rpi-firmware instead of our custom one, or at least more closely than we do now
18:13:02 <concrete_houses> why is firefox slow and sluggish when freebsd report 0.5 load avg?
18:13:09 <concrete_houses> swap 10g free
18:13:28 <concrete_houses> i am maxing upload network with torrents
18:13:33 <concrete_houses> but sometime sfox fast
18:13:48 <kevans> oh wait
18:13:48 <markmcb> kevans: for me it's the missing overlay for the rpi-poe-plus device (PoE+ Hat)
18:13:51 <kevans> I'm thinking of psci monitor
18:18:19 <concrete_houses> is there a free software version of single isgn on with keys?
18:44:46 <markmcb> if you're talking web, i use and like authelia
21:12:33 <Oleg> Guys, do you know why only "/dev/input/event0: System keyboard multiplexer" shows "events" when keys are typed, but "/dev/input/event5: Generic USB+PS2 Keyboard" doesn't show "events"?
21:12:38 <Oleg> for example,
21:13:40 <Oleg> Event: time 1689110005.359786, type 1 (EV_KEY), code 42 (KEY_LEFTSHIFT), value 1
21:14:11 <Oleg> when I use evtest, this type of information gets displayed for /dev/input/event0
21:14:40 <Oleg> but /dev/input/event5 has zero reaction.
21:16:21 <Oleg> I am asking this because when I use kodi on the bare console, the keyboard is fully unresponsive unless I delete /dev/input/event5. But even with /dev/input/event0, I have problems using the keyboard in the kodi player.
21:16:36 <Oleg> RhodiumToad: maybe you know something about it
21:18:30 <Oleg> If both /dev/input/event0 and /dev/input/event5 exist, the keyboard is completely unresponsive in the kodi player.
21:20:14 <Oleg> If I delete /dev/input/event5, kodi reacts to the keyboard, but often not in a completely appropriate way.
21:22:25 <Oleg> Is there anything I can do to make /dev/input/event5 to gain access to "events"?
21:23:26 <Oleg> "/dev/input/event6: DELL DELL USB Laser Mouse" doesn't have this kind of issue. it behaves appropriately in kodi.
21:27:53 <Oleg> folks?
21:29:27 <Oleg> Maybe I could edit devd.conf to make /dev/input/event5 my primary "event" device?
21:33:11 <Oleg> "cat /dev/input/event5" doesn't produce garbage characters when I type any key
21:33:24 <Oleg> I don't get it.
21:34:20 <Oleg> Why is the keyboard multiplexer (/dev/input/event0) responsive but the "generic usb+ps2 keyboard" isn't?
21:35:32 <Oleg> I think kodi specifically wants the "generic keyboard" to be responsive when it's run on the bare console.
21:42:28 <Oleg> I don't understand why /dev/input/event6 (the dell mouse) is an active device, but /dev/input/event5 (the generic usb keyboard) isn't.
21:47:36 <Oleg> https://www.gsp.com/cgi-bin/man.cgi?section=4&topic=kbdmux
21:47:37 <VimDiesel> Title: KBDMUX(4) - keyboard multiplexer
21:47:58 <Oleg> It says "The kbdmux keyboard driver provides support for basic keyboard multiplexing. It is built around the idea of a “super keyboard”. The kbdmux driver acts as a master keyboard consuming input from all slave keyboards attached to it."
21:48:35 <Oleg> so, /dev/input/event5 is not responsive because it's a "slave keyboard" device?
21:52:14 <Oleg> if I add hint.kbdmux.0.disabled="1" to /boot/device.hints, will the "slave" generic usb keyboard become a "master" keyboard?
21:52:29 <Oleg> or will it simply stop functioning?
21:53:22 <Oleg> clearly, there is something kodi doesn't like about the fact that kbdmux is a "master" keyboard.
21:55:57 <Oleg> I am gonna reboot with this hint.kbdmux.0.disabled="1" thing enabled to find out whether it will resolve my problem or make it worse.
22:10:58 <Oleg> I don't get it.
22:11:40 <Oleg> the kbdmux event device is gone, but the problem remains the same.
22:15:14 <Oleg> and now I can't even use the keyboard when running a wayland compositor.
22:24:32 <Oleg> okay, so I re-enabled the kbdmux device.
22:24:44 <Oleg> but I don't know what to do now.
22:35:54 <Oleg> I am out of ideas.
22:53:31 <concrete_houses> I am on hp z240 with icewm and firefox and life is good 13.2
22:53:46 <concrete_houses> is openbsd.org better as firewall n router?
23:13:21 <markmcb> all comes down to what you want to do. i tend to recommend OPNsense to people. freebsd core with a nice webui.
23:44:01 <foxiepaws> it really depends on what you like and know
23:44:26 <foxiepaws> opnsense is nice enough.
23:47:06 <foxiepaws> i've been using plain freebsd for years as my router, never had any issues with it. i like the version of pf on freebsd, and its my preferred firewall so i use that. All my other servers use freebsd as well so it helps me keep everything up to date and easy to manage because i'm not managing multiple OS