anyone know why firefox when I save a page does not do it right? I open page and its kinda wakked

Sorry, cannot parse

save web page

open saved page

kinda jumbled

looks not at all as orig page

firefox

You may want to choose the option of "Web page, complete" from the "Save page as" dialog

eshell

sorry

yea its all garbled

i didn't even know there is such a thing bugs.freebsd.org/bugzilla/show_bug.cgi?id=272731

Title: 272731 – Question on support booting 64-bit FreeBSD on a 64-bit CPU from a 32-bit EFI environment

well, it looks like it's not going to happen? :)

I'm looking to setup a FreeBSD NAT box again.

For years I was a dyed in the wool natd user. . . infact I wrote this doc which got a lot of downloads: trioptimum.com/~crtxreavr/quickanddirtynatd.txt

Anyways, I'm looking to use in-kernel nat this time.

I just read over docs.freebsd.org/en/books/handbook/firewalls/#in-kernel-nat

Title: Chapter 33. Firewalls | FreeBSD Documentation Portal

The handbook page seems a little vague on some of the filenames.

anyone knows what this value does in nfsd ? nfs_server_maxio (int) value to set vfs.nfsd.srvmaxio to, which is the maximum I/O size for the NFS server.

what's a maximum IO size ?

nvm, found it in the docs

A sysctl called vfs.nfsd.srvmaxio has been added that can be used to increase the NFS server’s maximum I/O size from 128Kbytes to any power of 2 up to 1Mbyte. It can only be set when the nfsd threads are not running, and will normally require an increase in kern.ipc.maxsockbuf to at least the value recommended by the console log message generated when setting vfs.nfsd.srvmaxio is first

attempted. 9fb6e613373c

how do I create a user for a service (i.e. nginx) without a home folder and disabled login?

Set the home to /nonexistent and shell to /usr/sbin/nologin

gist.github.com/Freaky/2560975d3c94246b86f464b8be75c967 Zenbleed mitigation rc script

Title: FreeBSD Zenbleed mitigation rc script · GitHub

I'm testing NFS performance on my local lan and I'm getting 400MB/sec with nfsv3 tcp vs 15MB/sec with nfsv3 udp

both mount parameters are the same except for tcp/udp option

why the huge difference ?

UDP packets are not going where they oughtta

?

sadly they are :) netstat -i shows no iface errors

just wanted to check whether conventional wisdom was that tcp or udp should be faster

why do you want udp nfs ?

I always thought it was udp

I want more speeeeeeeed :)

you should try nfs4

which is also tcp

yeah, but all benchmarks show nfs4 as slower than nfs3

How to kill a process despite in D state?

D as in 'in disk'? same as any other kill

rtprio, our modern perception of UDP is kinda backwards from its original intent.

UDP was intended for use on local, reliable networks.

But in the age of multimedia streaming, UDP has been utilized to reduce traffic (TCP checksumming) for high-volume traffic, likely crossing many routers.

So yes, NFS v2 & v3 default to UDP. They are also both SunRPC protocols, so they don't use a well-known port, they rely on sunrpc protocol to tell them which port NFS is running on.

NFSv4 has always been TCP-only, AFAIK.

Having disk issue; `smartctl -x /dev/ada0`  says ok  termbin.com/7r8w  but  `zpool clear ssd2` says `cannot clear errors for ssd2: I/O error`,  `zpool status` says  `state: SUSPENDED` and `ssd2 ONLINE 0 0 0` `ada0.eli ONLINE`

`errors: 229275 data errors, use '-v' for a list`

errors: List of errors unavailable: pool I/O is currently suspended

Beladona: i would guess, based on your formatitng, that you need to unlock the volume

Beladona, Can you pastebin the "zpool status ssd2" output? And also the .eli implies that it is a geli encrypted volume below the zpool is that right? I think (not sure) that rtprio's comment was about the geli status of that volume and I agree I would verify that the volume is unlocked/decrypted for zfs to access.

Maybe look at "geli list" and verify that ada0 and ada0.eli are properly consumers and providers and decrypted okay?

CrtxReavr: interesting; i should have known that, but i suppose ip/tcp/udp use has changed quite a bit now from the 70's and 80s

rwp rtprio true, its geli based but was unlocked and I was writing a postgres dump on it already when this happened

so sort out the geli and it should start working

unless you already have and now you are trying to get zfs to recognize it

rwp Geom name: ada0.eli  State: ACTIVE

rwp rtprio   geli attach /dev/ada0  Enter passphrase:  geli: Cannot access ada0 (error=1). geli: There was an error with at least one provider.

That does look to be the problem. At the geli layer. Once that is resolved then zfs should be able to use it.

rwp how to resolve it and whats wrong/

no logs imgur.com/MvsEQdU.png

"I too like to live dangerously!"

Unfortunately I am not using geli encryption for anything other then temporarily encrypted swap in the standard way from the installed and do not know how to debug geli encryption problems myself. But hopefully others might know.

However the error message you report is "cannot access ada0 (error=1)." so I would start there.

Among other things I know the OS removes devices from /dev/* after the alternative names are opened. To enforce singleton access.

I am still curious as to what "zpool status" shows. And if not that then possibly what "zpool import" shows. Possibly also "geom -t". Those all list the detail structure of the storage.

ok

CrtxReavr, I need to go laptop shopping. The battery in my current lightweight everyday carry machine has failed almost completely rather suddenly in the last month. And a replacement battery is currently unavailable.

CrtxReavr: dangerous aside, you must have some patience for that

i considered getting a lenovo x1 keyboard but not sure i can wait that long

I know of no other option.

Well, this is for an X1 Carbon. . .

They keyboard died, found a replacment, easily and quickly enough. . .

But then I broke the touchpad ribbon cable clip when I went to install it, 'cause I didn't properly understand how it opened.

And AliExpress is the only source I know for those clips.

Howdy. It seems on FreeBSD one does not set specific network protocol parameters with sysctlt, as you do on Linux?

Depends what it is.

sysctl was stolen from FreeBSD. . . just say'n.

rwp # zpool import     no pools available to import

# zpool import ssd2           cannot import 'ssd2': a pool with that name already exists

CrtxReavr: Do u trust on AliExpress?? Min. order 10000 :(

Beladona: zpool import -f

I needed ~$16 in parts.

tyler82, didnt' trust them enough to give them a CC# - paid with paypal.

tyler82 same

Beladona, No pools to import means it does not find any providers available. Which corresponds to the geli failing to decrypt. So far that is consistent.

ya

Beladona, you didn't need that migit amputee porn anyways.

CrtxReavr wihtout it theres no fun in life

CrtxReavr, I'm searching for the Linux equivalent of `sysctl -w net.sctp.encap_port=10`. It seems nothing net-related exists on FreeBSD's version

just kidding

anyone here with good knowledge of freebsd audio support? i mean internals. i may try mailing list but do not know which one exactly

audio, video, hardware drivers.. freebsd? different thigns

these are the things I messed most with

not easy

net.inet.sctp.udp_tunneling_port

I am trying to figure out whats the best solution for full system backup. On linux i use timeshift but on bsd....i donno. I was trying some programs like Amanda Borg Restic but all has got issue with some file ownership...even if i run as root. The only thing is really working for me rsync or rsnapshots. Combine with zfs snapshot and zfs send with tar bzip2.

i do tar+rsync

have been meaning to try bacula

bzip2 sounds painfully slow.

Beladona, Everything I know about geli I learned from docs.freebsd.org/en/books/handbook/disks/#disks-encrypting-geli

Title: Chapter 20. Storage | FreeBSD Documentation Portal

gzip should be faster and compress adequately.

rtprio: yeah. sounds a reliable solution for me

rtprio: i also tried bacula. not my fav.

CrtxReavr: but bzip2 has better compression rate. but slower and more source hungry.

rwp I still wonder. If the disk is gone, then reboot wont matter here

CrtxReavr, where did you find that?

tyler82, from my experience, the marginal extra compression isn't worth the time or CPU cycles.

Also, there's pigz. . . a mult-threaded, yet thread-safe implementation of gzip.

jemius, I don't know anything about encap_port=10 but on FreeBSD most networking is configured using ifconfig+route and the init scripts configure this using /etc/rc.conf variables. I would read the docs for ifconfig and see if that shows it.

jemius, sysctl net.inet.sctp | grep -e encap -e port

ok, i will ask. maybe anyone with good knowledge can look. the weird issue i have and which i consider to be really unusual ( and possibly bug ) is: i have intel onboard audio card ( dev.pcm4 it is ). i have set bitperfect mode and also set vchans to zero for it ( no multiple channels possible ) and it works just fine. now that i have xfce4 as desktop i also have pulseaudio running automatically and for unknown reason it can play

multiple sources ( say music and youtube video ) simultaneously. this is not how bitperfect mode supposed to be with mixer channels disabled. unless i really missing something

Isn't there somewhere a manpage for each sysctl node?

hmm. thx!

jemius, I'm not positive that's what you're looking for.

Beladona, You are not booting from ZFS? It's just data? Then a reboot _probably_ will not matter. But probably won't help either. (shrug) I don't know because we don't know why you are getting error messages trying to decrypt that geli provider.

ok

jemius, I would start here docs.freebsd.org/en/books/handbook/advanced-networking for a basis and then ifconfig specifically. What does encap_port=10 do?

Title: Chapter 34. Advanced Networking | FreeBSD Documentation Portal

CrtxReavr, why not?

rwp, There's this ancient protocol SCTP. Many routers block it. That's why you can tell the kernel to encapsulate it into UDP. You have to activate that and maybe specify a port number for it (10 in this example)

It's not ancient. . .

It was written as a drop-in replacment for TCP, in the mobile age.

Ancient is the wrong word. Let's say middle-box-plagued ;)

TCP tracks a "connection" with two ip:port sockets.

SCTP (Stream Control Transport Protocol) uses a higher level session identifier that allow for a session to be maintained with an IP address change at either end.

Thanks for the summary. I have not encountered that before.

I learned about it in an IPv6 class my $work paid for.

And yes, It does stupport LegacyIP.

(IPv4)

I know quite a bit about SCTP. And it might support v4, but no router will route your stuff. v6 works quite well in many networks, though, because there's no NAT

anyways, worked. I'm happy :)

jemius, "worked"? What did you do that worked? So that I might be educated? :-)

rwp, I just wiresharked my echo-server to check it's actually encapsulated

So it was already automatically encapsulated without your needing to configuration anything?

Next I'll check if it's routed through my country's LTE-Provider. They are blackholing pure SCTP due to unknown reasons

rwp, no, it gets encapsulated when you set the encap-port

via sysctl

Additionally, it's Linux interoperating with FreeBSD here. It's the same protocols with the same RFCs, but you never know...

What was the sysctl you used to set it? ELI5? No maybe 7! :-)

sysctl net.inet.sctp.udp=9899 <- see above

*net.inet.sctp.udp

Okay. (I don't see that string in my search for it in the scrollback.) Thanks!