What's the right way to escape a modifier in bsd make?

a backslash _xor?

`man 1 make`

Yes that's what I tried first and I did read the man page (looking at it again now, in fact). It didn't work, so there's something specific about this usage of it that I'm missing.

I'm finishing up a port. It builds fine, but does throw a warning.

brb

Anyone know Why I would be getting the version mismatch from pkg upgrade like this:

- package: 1301000

- running kernel: 1300139

Even though it is at 13.2-p2 ?

This is in a jail.

If I run freebsd-version -kru on the host they are all 13.1 p2

If I run it in the jails they are also 13.1 p2

(NO kernel of course in the jails)

And above 13.2 is a typo

pkg upgrade worked fine on the host

It is strange right?

I could do IGNORE_OSVERSION=yes

But I wasn't sure what my problem was

Looks like it. Perhaps missed some step during OS update?

Maybe

IGNORE_OSVERSION=yes seems to have worked

I am going to reboot for the heck of it

Restart all the jails.

And then try pkg upgrade

I noticed I had missed this server, and it had not been upgraded to 13.1 until today

So when I first ran pkg upgrade it did have the wrong version

Maybe it stored that flag

Seems to be running smoothly now

Weird...

Post reboot(s)?

Well, rebooting did not fix that error I had to do the IGNORE_OSVERSION

But after setting that and running pkg upgrade and rebooting it works.

Hmm ...

meena: FYI 13.0-RELEASE reached end of life a few weeks ago, <freebsd.org/security/unsupported>

Title: Unsupported FreeBSD Releases | The FreeBSD Project

Does any one have experience running FreeBSD on HP Proliant towers? I bought a used ML30 Gen9, am I going to experience pain trying to use third-party components (HBA, drives etc)

Since you have already bought it one must wonder if it matters? Since regardless you will almost certainly try it anyway. And then you will know! :-)

Experience via fire ;->

I am running FreeBSD on previous generation HP Workstations. Typing this on one right now.

The only problem I have has been with some but not all HP Workstation UEFI firmware. On five systems close in time but not identical ONE of them will not boot UEFI without crashing during the boot.

One would like to think that these bugs were found and fixed in later generation models. So you are probably going to like the machine very much.

Which CPU did you get with that model? Hopefully the blazing fast E3-1280v6! But then I would be super envious. :-)

rwp: yes, I will try it >_<

It has a single E3-1230

It's got a SFF 8-bay drive cage, but I don't know if I'll use that or swap it for the 4-bay LFF

I need to buy the disk caddies either way

Just heard stories about the fans ramping up to 100% when non-HP drives are used, which does not excite me

I think that one with the E3-1230 only ever shipped with the E3-1230v6. It's quite speedy! cpubenchmark.net/cpu.php?cpu=Intel+Xeon+E3-1230+v6+%40+3.50GHz

Title: PassMark - Intel Xeon E3-1230 v6 @ 3.50GHz - Price performance comparison

I'll need to double check, but the specs on ebay said v5 3.4GHz

Does HP uses others' disks but branded or does it produce its own disks?

s/uses/use/

In any case, the plan is to get an LSI HBA to replace the Raid controller, a PCI m.2 card and drive for OS, and attempt to use consumer disks in the 8-bay

parv: no idea actually

tehpeh, Ok

HP hasn't made their own disks in a long time now. But they did at one time.

Interesting

The diagram here is informative of current vendors. en.wikipedia.org/wiki/List_of_defunct_hard_disk_manufacturers

Title: List of defunct hard disk manufacturers - Wikipedia

This site indicates that some drives are ok, and some are not (at least for an DL380) dascomputerconsultants.com/HPCompaqServerDrives.htm

Title: HP Server Third Party Drive Compatability

tehpeh, Earlier you mentioned switching 8-drive bay cage with 4. What would you have planned in either case?

grahamperrin: need to shuffle order on PkgBase.live, and upgrade my own jails :😬

:-)

On that page it is discussing HP Proliant DL380 servers. That's a rack mount system. The HP tower you just got will have been from a different design lab.

parv: not 100% sure. The 8-bay takes 2.5in SAS (and hopefully SATA) drives. One idea I had was using the 5TB Seagate portable harddrives, cracking them open, and using the drive inside. Cheap, but unsure if it will cause issues with ZFS.

meena: … and maybe encourage probonopd to take more of an interest in PkgBase

rwp: good to know! It makes sense that the Proliant Microserver would take any drive because it's marketed at consumers really, and the ML30 is only one step up from that...

I don't need a lot of storage, 2TB minimum, 5TB would be nice, in a mirror

Don't forget the backups!

Oh yeh, good point, might do that on a separate system though :D

If I swapped for the 4-bay I'd buy new 3.5in NAS drives

.oO( Man, only of 8-bay thing could have 3.5 in disks )

s/of/if/

tehpeh, I have posted this here before and risk being too repetitive but if you are buying drives now be aware of the problems documented here raid.wiki.kernel.org/index.php/Timeout_Mismatch

Title: Timeout Mismatch - Linux Raid Wiki

Yes, exactly! I only found out about this yesterday, very interesting

And avoid SMR drives! Just recently someone else here got caught in the SMR trap.

Which means using enterprise 2.5in (probably SAS) or if I can get a 4-bay cage I can use cheaper 3.5in NAS drives (I believe)

Or just wait until SSDs become cheaper for 2TB+

It all depends upon how much storage you want online. For a 6x 6TB array spinning is still a LOT less expensive than SSD for example.

But a 2x 1TB SSD pair is certainly cheap and affordable.

On SMR, avoid WD "Red"; "Red Pro" are ok; don't know the deal with "Red Plus"

parv, Look at the map key here: blog.westerndigital.com/wd-red-nas-drives

Title: On WD Red NAS Drives - Western Digital Corporate Blog

crazy that WD would see a drive branded NAS that wouldn't work properly in RAID

WDC's management thought shipping cheaper SMR drives as NAS RAID drives was a good thing. Got caught. Got shamed. They came clean and said here is how you can tell what you are actually buying.

Fair, but they should also print that on the box ;)

Meanwhile Toshiba used that as a marketing opportunity to say they had never lied to us. toshiba.semicon-storage.com/ap-en/c…ics/2020/04/storage-20200428-1.html

Title: Use of Shingled Magnetic Recording (SMR) technology in Toshiba Consumer Hard Drives. | Toshiba Electronic Devices & Storage Corporation | Asia-English

As long as I am being repetitive in the channel go to page 2 here and look at the resilvering times. servethehome.com/wd-red-smr-vs-cmr-tested-avoid-red-smr

Title: WD Red SMR vs CMR Tested Avoid Red SMR - ServeTheHome

I just think information is power and don't want people to get caught unawares!

Ok, so I'm going to target 5TB usable space in a ZMIRROR (so 2x 5TB or 4x 2TB maybe) and the 2.5 vs 3.5in will be decided by cost I guess

rwp: nice, thanks for those links, will read tonight

And I don't know if I want to advocate for this here but... It's somewhat an interesting opinion piece. So here it is but it is BUYER BEWARE for certain! "Cost of SAS vs SATA Hard Drives | Shopping with Art of Server on eBay!!" youtube.com/watch?v=QtvJA9mHNjw

Title: Cost of SAS vs SATA Hard Drives | Shopping with Art of Server on eBay!! - YouTube

thanks both rwp and parv :)

tehpeh, Aye! Bye!

tehpeh, Sorry, it was rwp who was relocating.

hehe

😅

So what happens to the people who buy synology or qnap NAS boxes and put SMR drives in them?

tehpeh, What do you mean? Technically (ZFS doesn't like) or from consumer point of view?

Well presumibly those consumer NAS boxes don't run ZFS, so are they at risk?

Don't know enough; neither have I read an anecdote that I could rememer

s/rememer/remember/

Just curious :) Alright I'm off too, have a good day y'all

Later

Is there an equivalent to CAP_NET_BIND_SERVICE in BSD?

MAC or Capsicum may provide one. Hard to say without knowing what it does.

Allows (induvidual binaries) to bind <1024

Looks like it might just be sysctl security.mac.portacl.rules=uid:80:tcp:80,uid:80:tcp:443 (etc) doesnt help the example uses uid 80 took a few reads to get over port 80

I think you can do that via MAC, but it's a pretty steep curve to climb, for something you can just declare in a systemd unit file or or an SMF service definition

FatalFUUU: and that's then for a user, and not restricted to the lifetime of a process / service

not sure I follow you meena - where would systemd enter into this ?

FatalFUUU: systemd offers a config option, freedesktop.org/software/systemd/ma…temd.exec.html#AmbientCapabilities=

Title: systemd.exec

in bsd?

it would be cool if our rc did the same

I know about this just not sure why it was brought up

New laptop running 13.1. I have an Intel wireless 3165 PCI(mumble) device that's recognized by iwm(4), but the firmware for that is missing, and freebsd.org/status/report-2022-01-2…linuxkpi_802_11_compatibility_layer (and erlier status reports), wiki.freebsd.org/WiFi/Iwlwifi/Chipsets, and the 13.1 release notes all say or hint that I

Title: FreeBSD Quarterly Status Report First Quarter 2022 | The FreeBSD Project

should. But iwlwifi is loaded by devmatch and iwm appears to be loaded much earlier, in loader or when probing for devices. How do I change that? iwm_load="NO" in loader.conf? device.hints looks ISA-only.

s/that I should/that I should use iwlwifi instead/

Oh, module_blacklist. Going to try that.

Booting from a simple installer for FreeBSD on a USB flash drive, is the command below enough to (intentionally) delete everything on the internal nvd drive?

trim -f /dev/nvd0

hm, wouldn't handing the socket to listen on to the service be a better idea, but rc probably can't do that as a sh script, maybe add something like that to daemon?

apparently, openntpd does not give a toss about how much drift there is when it starts up

it does not just panic, and instead just adjusts time

you set the -q flag?

FatalFUUU: mostly thinking out loud. I like to pick on init systems, and think about of ways to improve them

s/-q/-g/

nimaje: that's not enough, either

nimaje: ah, -g should be, yeah,

but, no, i wasn't setting any flags on either daemon. openntpd works without flags

yeah, just assumed that openntpd would behave like ntpd for that, before looking it up, seems like openntpd ignores how big the adjustments are and just makes them

Why is that -- openntpd does not give a toss about how much drift ... just adjusts time -- a bad thing?

it's a hole for attacks

Or, what would be gained if openntpd would refuse to adjust time on "too large" (for some definition) of a drift?

because something is very wrong if your clock drifts too much while running (well, or someone tries to attack you for some reason and tries to set your clock to a specific time)

my (ports) system fucked itself suddenly - any way to debug it?

what are the symptoms?

accounts-daemon, lightdm and polkitd all segfault

hm, where do they segfault? lldb can help you with that; maybe a partial update?

vxorps

is that avx? I don't have avx

oh I see what's going on

nimaje: the main reason my clocks drift that much in my VMs is because I put my laptop to sleep

what the heck happened to thunar?

yeah, but something like that is probably the only way to legitimly have big clock drift on a "running" system

one might wish that the hardware RTC ran during a sleep, and that it was read for a time-jump on wakeup, and that time-sensitive applications (eg. ntpd) got notified of wakeups.... I don't know if any of that is true. Time is hard.

but the case here was a vm and the host is put to sleep, so the vm thinks no time passed while the host was sleeping

anyone done the UFS snapshot as Boot Env dance yet?

so host should pass on wakeup notifications to VMs

jgh: virtio has clocks, but i know that OpenBSD doesn't implement that, and last time I read around that corner of code, neither does FreeBSD

hrm, i'm seeing an RNG in libvirt, but no clock… dunno if that means it's happening automatically, or that I'm wrong

(me, being wrong is quite common tho)

when will they fix webkit in 12.3?

hhfyu: What's wrong with it?

it is missing

What exactly is missing from where?

webkit package is missing from the repository

on portsfallout.com for webkit I only see that qt5-webkit doesn't build for riscv64 as it doesn't know the cpu

Title: FreshPorts -- www/webkit2-gtk3: Opensource browser engine using the GTK+ 3 toolkit

from which repository, there are many as you need one for each pair of freebsd version and architecture

12.3, x86_64 obviously

ah, and quarterly or latest

how ca I tell

half of xfce desktop is missing due to a stupid epub preview library lol

probably quarterly then, as that is the default, you can configure that in the config files for your pkg repo (see url there)

the latest build of 12.3, quarterly, amd64 that I can find has a successful build of webkit2-gtk3 (and -gtk4), so no idea why you don't get it

I see gtk4 one but not gtk3

last year there was talk about panfrost being "almost ported"

but since then, there's been no info about it and seemingly nothing is happenning

does anyone know the current status?

package system in freebsd is laughably bad

ORLY?

from non-existent infrastructure through dependency handling to packages vanishing

For desktop / non-power usage: Generally speaking, mixing large-to-medium scale ports with pre-packaged binaries is going to be the cause of one's real headaches. Ports =/ AUR. But the pre-packaged quarterly binaries are rock-solid, and an occasional small port will be good.

there is half of xfce missing, in your rock-solid packages

aur is much more usable than ports

WTF is up with mergemaster?

It's just showing me the diff, it's not letting merge.

ok, no idea which pkg repo you use; as far as I can see beefy2 should build 12.3, quarterly, amd64, the latest build there is beefy2.nyi.freebsd.org/build.html?m…3amd64-quarterly&build=31229ca1324c there xfce succeded

hhfyu: so you use freebsd 12.3 on amd64 and have url: "pkg+http://pkg.FreeBSD.org/${ABI}/quarterly" in /etc/pkg/FreeBSD.conf and don't override that somewhere (e.g. in /usr/local/etc/pkg/repos/)? pkg.FreeBSD.org uses geo-dns and tells you which mirror you get if you just visit it via a webbrowser

yes

browser says 403 forbidden

the beefy link doesn't open either

hhfyu: imgur.com/aWtHd74.png

are you sure you have properly set up network?

ok, for some reason on my mirror pkg.freebsd.org/FreeBSD:12:amd64/quarterly has modification dates of 2022-10-01, while the quarterly was still running 2022-10-02

Title: Index of /FreeBSD:12:amd64/quarterly/

hm, I throught beefy < 10 support ipv6 and ipv4, but it seems like beefy2 only has an ipv6 address

I think most of build servers are IPv6 only nowadays, but they're accessible through pkg-status.freebsd.org by using pkg-status.freebsd.org/beefy18 (as an example)

Title: Poudriere Index

Or you can ssh -D 1080 to a server with IPv6 and configure your browser to use a SOCKS5 proxy on localhost:1080 and send DNS queries over the proxy.

s/server with IPv6/dual-stacked server/

meena: from pciconf(8) DESCRIPTION <freebsd.org/cgi/man.cgi?query=pcico…ktion=8&manpath=FreeBSD#DESCRIPTION>,

Title: pciconf(8)

> … The first column gives the driver name, …

grahamperrin: i found devinfo to give me the answer easier tho

The page was a wall of words until V_PauAmma_V helped me to understand the significance

grahamperrin: devinfo -p vtnet0 ⇒ vtnet0 virtio_pci0 pci0 pcib0 acpi0 nexus0

I stumbled across the same thing a day or so ago, maybe we were in the same place. Sleep!

grahamperrin, I did? You meanlive-IRCing my fumblings helped someone else instead of getting me help? :-)

s/mean/mean /

V_PauAmma_V works in mysterious ways

Mysterious to myself as well, looks like. :-)

;-]

V_PauAmma_V: it's the type of thinking aloud that (often) helps me