is the syntax for ethernet interface alias in /etc/rc.conf still ifconfig_bge0_alias0="inet 99.104.201.113/29"

really odd it won't boot with the alias but if I do service netif restart then there it is

crb__, I think you are missing an alias keyword there. I think it should be: ifconfig_bge0_alias0="inet alias 99.104.201.113/29"

But you are right. Looking at man rc.conf and then /alias they don't list it in the command. But that is what I have on my system.

Looking at "man ifconfig" the syntax listed there is that alias appears after the address? (me shakes my head) I should read the docs more often.

The example in the man page is "ifconfig em0 inet 192.0.2.45/28 alias". Hmm...

It seems that "alias" can work in either place before or after. But "-alias" to remove it can only work in the after position, as documented.

jbo: you ever create a port that you look at after the fact and think, "man. this shit is utterly cursed."

I have a prototype arm-none-eabi-libstdc++ port. it starts to build. no idea if it builds too much. no idea if it builds what we need, yet.

we'll see what poudriere thinks about the plist once it's done

I think the way my rc.conf file worked with the "alias" in there was by accident! This was a good opportunity for me to figure that out and fix it.

rwp, nope alias before doesn't work

rwp and adding it at the end didn't help either

as I said very strange that simply running "service netif restart" with once it boots causes it to show up

I am sorry crb__ but as you read I was wrong about the syntax. That was the syntax I had in my /etc/rc.conf file for it. But when I read the docs my extra "alias" there was clearly wrong.

I removed it from my rc.conf file to correct it here. So... Thanks I guess for causing me to figure out I had it wrong here? :-)

crb__, It is strange what you report. But there may be some difference between start and restart later.

Have you tried stopping the service and verifying with ifconfig that everything is offline. Then running start from a cold state?

I might even go so far as to comment out networking entirely and rebooting to get to a pristine start state and then uncommenting the network config and running start from the cold pristine state. Maybe a failure there? I have seen problems like that before.

Also, is there any clue in /var/log/messages about a problem? Or in the dmesg output? I would expect an error to be logged somewhere.

what's this about interface aliases?

gh00p, crb__ is reporting a strange problem where "< crb__> really odd it won't boot with the alias but if I do service netif restart then there it is"

And the configuration under review is: ifconfig_bge0_alias0="inet 99.104.201.113/29"

Assuming the interface name is bge0 then it seems okay to me.

This is all documented in "man rc.conf" look for "alias" there.

rwp, you don't need 'inet' in there.

rwp, an `ifconfig_bge0_alias0` line should not have the "alias" keyword included in the value.

gh00p, nonsense.

ifconfig(1) does however say:

CrtxReavr: mine works without. I haven't modified my startup scripts.

a non-conflicting netmask must be given. Usually 0xffffffff is most appropriate.

Now. . . I can remember a time when that was absolutely true for FreeBBD.

At somepoint it seemed to stop being a requirement for aliases, but there may be corner cases.

rwp, try: ifconfig_bge0_alias0="99.104.201.113/32"

(Assuming you have another ifconfig_bge0 flag with a proper mask width.

)

CrtxReavr, Don't tell me. I am just the messenger at this moment. But the man rc.conf says it is needed.

CrtxReavr: "nonsense"? You don't have the "alias" keyword in your value either.

gh00p, I missread what you wrote.

Thought you were claiming the opposite.

ah

gh00p, Where do you see that crb__ has an alias in the value?

rwp, don't look at rc.conf(5), just look at /etc/defaults/rc.conf

rwp it was your suggested line maybe 3 hours ago

So, I just want to make sure this is down in writing. You are admonishing me to NOT read the fine manual. Correct? :-)

gh00p, Yes 3 hours ago, and 2.9 hours ago I said I was wrong about that and posted a correction.

it started, "I think you are missing an alias keyword there."

it started, "I think you are missing an alias keyword there."

But in the half hour ago time I just summaried an answer to your question "< gh00p> what's this about interface aliases?"

oop

not everything was entirely clear...

rwp, well, experience and /etc/defaults/rc.conf say one thing and rc.conf(5) says another.

It was a dead reconning navigation error due to missing data in the middle.

time for a PR? :)

So there are two examples in /etc/defaults/rc.conf and neither really apply but the closest example there has "inet" in it with "ifconfig_lo0_alias0="inet 127.0.0.254 netmask 0xffffffff" # Sample alias entry." Though that is for the loopback device it has the same syntax.

The other ex

The other example is "ifconfig_em0_alias0="inet6 2001:db8:2::1 prefixlen 64" # Sample IPv6 alias" which is for IPv6 of course and so differs accordingly.

inet is the default, so it's not necessary.

I don't really see the conflict between /etc/defaults/rc.conf and the rc.conf man page.

Not being necessary is not the same as being incorrect or invalid to be there.

And with v6 ifconfig_em0_ipv6 makes specifying 'inet6' needless.

I also think that if you believe it to be incorrect then please file a PR so that it can be discussed there and corrections taken if it is needed.

I've been beating that drum in #bsddocs on EFnet for ages.

I mean my only part here is that I am referencing the fine documentation and if you think the docs are wrong then argue that out with the people who have commit access to the docs. (Which I do not.)

Given what I see in /etc/defaults/rc.conf the two examples there converge with what man rc.conf documents. I think they are in sync.

Also the man page for "man ifconfig" has this example in it "ifconfig em0 inet 192.0.2.45/28 alias" so the ifconfig man page also documents using "inet" there.

gh00p, BTW... I started walking through the rc script to see why my config with the extraneous "alias" in it was working. But it was there, and it was working, but I didn't think it should. I became distracted before I got to the conclusion but I think that "alias" became an ifconfig null action with no address before it and then ifconfig continued processing the rest of the value from there.

Accidentally working by happenstance is not a good thing so I was greatful that crb__'s question got me looking at my own config and got me to find a problem I had and fix it.

My next step in the root cause understanding would be to look through the ifconfig.c source and see how it is parsing its command line. I am curious how that is implemented.

But that will have to wait because I am now relocating. BBIAB. Actually probably a few hours at best and tomorrow morning more likely.

jmpp: thanks for explaining it

tercaL: anytime!

I'm currently on the finishing touches of a major overhaul of my home tech stack, and once I finish I'll get back to uptime-kuma to polish it

and, if I manage, to submit it as a port

but I aint making no promises :P

jmpp: Oh, good luck enjoy your time!

tercaL: jails distributed across purpose-specific VLAN's, it's been loads of fun!

jmpp: By the way, while I'm pretty new to Node.js, would you recommend to have a separated user (either 'www' or a freshly created one) for any node.js application or this is a case specifically for uptime-kuma?

uptime-kuma is the first node.js application I ever installed/tested.

I'd gone for www:www, just out of simplicity, as any other web app, regardless of its backend stack

hmmm..noted down.

and if you deploy one webapp per jail, that makes sense

so that's what I'll aim for in my uptime-kuma jail once I get to it, with customization hooks in the rc script in case anyone wants to change the defaults, of course

but, that being said, Morpheus is calling me!

jmpp: And whenever you need to open the service from jail to a public IP, I can -if you of course need- help you in PF, NAT and port forwarding configuration, just for you to know. :)

G'night!

tercaL: thank you, most appreciated! (though for the time being I don't plan on exposing uptime-kuma outside of my home network, and the VLAN the jail is on is heavily protected by pfSense)

Super fine, then!

still, good to know you're pf savvy, I'm heavily dabbling with that as well

and now, for real, have a good night ;)

jmpp: PF sometimes might be pretty useful especially when we're on bare-metal dedicated server (limited to a single public IP) and working with Jails

Good night!

is anyone running 15.0 on an rpi4? mine won't boot from SD card with the error "sdhci_bcm0: DMA error 4 on CH5" - but the card itself is fine, it works in the bootloader and it works if i boot from a USB card reader instead. tested with 4cd0ce41bb7e from yesterday

it also worked fine ~1-2 weeks ago when i installed it with a 15.0 snapshot

ok interesting, it boots with GENERIC, it doesn't boot with GENERIC-NODEBUG

O_o

lw: that sounds very interesting, indeed!

meena: i did a bug bugs.freebsd.org/bugzilla/show_bug.cgi?id=277884

Title: 277884 – RPi4: mmc broken with GENERIC-NODEBUG

lw: did you try GENERIC-MMCCAM?

no, but i will do that now

(i was trying to get uplcom(4) console working using hw.usb.ucom.device_mode_console but it doesn't seem to want to)

meena: GENERIC-MMCCAM or GENERIC-MMCCAM-NODEBUG? i guess i'll try both

My IRC client says I have a mention but I can't find it. Anybody pinged me?

cracauer, I see no mention of your nick in this channel since you and I talked about the radeonkms driver in ports and the kernel compatibility problems of kernel drivers in ports on Tuesday.

/whois VimDiesel

oops

why is there no if_genet.ko?

looks like the driver itself has module support, it's just not hooked into the build system

polarian, VimDiesel is the resident channel bot. wiki.freebsd.org/IRC/BotBrainstorm

Title: IRC/BotBrainstorm - FreeBSD Wiki

polarian: an actor on endless repeat of the same theme... over, and over, and over!

:P

how do I ask pkg to tell me what package update will cause the installation of a new dependency?

I just ran pkg update -f && pkg upgrade -n, and it told me that tcl86 will be brought in, and I cannot begin to understand why, what package has a newly grown dependency on tcl

I know I could do pkg info -r tcl86 after performing the upgrade...

but I'd love to know *before* I do so

I mean, before I perform the upgrade

well, I can only imagine that tcl86 will be brought in as a dependency of sqlite3-tcl, which I can also infer will be brought in as a newly grown depdendency of sqlite3

but that's just inference, there must be a way of programmatically inquiring that information

jmpp, Maybe this? pkg search -Q required-by tcl86

that just gives me the name of the package itself :(

pkg search -Q required-by sqlite3-tcl --> sqlite3-tcl-3.45.1,1 (plus the package comment)

sqlite3-tcl is the new package that's going to be installed

and I'm trying to inquire why

so I opted to do pkg upgrade -n $pkg for every package pending an upgrade... but, strangely enough, all of them seem to be bringing this new one in...? me confused!

When I run that for tcl86 it lists tcl86-sql as a dependency but then listing it says none.

Listing depends for tcl86 lists mysql80-client which of course has a longer list. But none of those sqlite3.

this is in a php jail, so I have php82-pdo_mysql installed, which would explain it

however, I'm not getting much output, strange

pkg updating -f raw.githubusercontent.com/freebsd/freebsd-ports/main/UPDATING --> unable to open, me sad :(

Hmm... bsd.to pastebin is offline again.

a couple of days ago I had to use mozilla's pastebin

Good luck! I must relocate.

rwp: thanks for your help!

hihi! i've been experiencing audio trouble with chromium and a usb audio device. i have a mic connected to a universal audio vault 1 and the mic input works great for a while -- until it it starts to sound robotic to listeners. this is under -current.

reloading the tab in whatever video conferencing software i'm using temporarily fixes (it is consistent across zoom, microsoft teams, and google meet).

i'm running sndiod and i believe chromeium is configured by default to use soundio.

rwp: I realised when I whois'd

the username seemed familiar

jmpp: what do you mean?

Vin Diesel :D

ayan: although this might not be related, you can try disabling the automatic mic gain feature in chrome*: In chrome://flags set "Allow WebRTC to adjust the input volume" to Disabled.

Just to double check, theres no port for jdtls right (Eclipse Java LSP server)

unless it is within the eclipse port... but I doubt it because eclipse likely bunded it in their IDE instead of it being its own standalone thing that jdtls is mean't to be

I like to install service-names-port-numbers in /usr/share/misc for this reason, it seems in the spirit of the directory. nothing comes up when I grep it.

my memory seems like we had it in there when I was a child but I can't find anything to support that

oops, i see i've misunderstood the question

So, I've got this Sonoff Zigbee interface that looks like /dev/umodem0 to FreeBSD. Is there software for these things, or is one meant to roll one's own? Can I run ser2net and access it from a linux homeassistant instance?