will smalltalk make it to freebsd anytime soon?

any tips on speeding up chrome?

concrete_houses, GNU smalltalk-3.2.5_17 is in FreeBSD 13.1R. Simply install it.

Much of the slowness of chrome/chromium/firefox other browsers is the latency of the Internet.

Also squeak-4.10.2_11 is available too. Perhaps the better choice even.

hmmm

that edselled

is there a good way to use protect(1) on the processes that will be started by poudriere? it seems like i should identify the relevant parent build process in each jail that poudriere starts, and use protect -ip $PID

jmnbtslsQE, How will keeping the process from being killed on out of memory conditions help? How can it continue?

Seems like if you need more memory for a task then you need more memory.

not sure, i want to at least give it a try. i think it should be able to find memory elsewhere, such as ARC, or from 'Inact' memory, and that this kill scenario is exceptional

going to try protecting them : ps ax -o pid,jail,command |grep -E '[b]uild_pkg' | awk '{print $1;}' onmy next build

I also hear plural build jobs. It might also be necessary to reduce the number of simultaneous jobs to reduce the overall memory needed.

Also if you want to keep pushing on then add more swap space. Swap space not on ZFS but on, say, another disk.

Just pop a single SSD in there and make it all swap. Something like that would help and would prevent processes being killed for being out of memory.

yeah, have lots of fast swap but not much being used (not sure how much used during the kill, but surely not most of it). good point about the number of jobs may need to reduce that

wanted multiple jobs for better utilisation of CPU/disk

hmm, may have been partly due to some periodic scripts that i can turn off

so happy I switched to UFS from ZFS

my smaller hp 610 works way better

63$ desktop mini

4tb external usb3 hd

lovin it

freebsd rocks

sounds and xorg just work

tight!!

archlinux main competitor havent tried void or alpine linux

musl instead fo glibc intersting slant

nice use of the enter key there.

anyone running quasselcore + postgresql in jails? interested in setup

I don't know what quesselcore is, but recently moved a few big dbs to jails when moving to new hardware...

quasselcore is an IRC client daemon (not quite a bouncer)

this is i know

I'm explaining for spork_css

depending on how you install your jails, the setup may be the same as if you are installing them in the root jail

("root jail" is my personal terminology for what's probably best known as prison0, or just the unjailed system)

AmyMalik: ah, sorry :)

AWS recently launched eu-south-2 region (Spain). Does anyone here know if EC2 AMIs will be automatically created next quarter, or if it's something I should open a ticket for somewhere?

shiroyasha: it's available now? find @cperciva on Twitter :)

why is it eu-south and not like es-whatever

AmyMalik: because it's <continent>-<region>-<number>

Espana isn't a continent

us isn't a continent either

corporations need to start making sense if they do not want to be abolished

the world is people, and people dont' make sense

if corporations make sense, they will stop being compatible with people

meh

hello, I would like to write C daemon watching log file and make some action if something occurs. I know about fail2ban but i want to learn programming in C so it's a self educational project. Any hints what could help? Thank you

I guess you should take a look at kqueue

for portability to other systems also epoll, select, and poll

mage, great, I'll look at it

AmyMalik, i want it as simple as possible so freebsd only

really, you need to be upstream of writing the logs to disk. this isn't an easy task under freebsd and other systems that liberally use syslog.

shiroyasha: I'd ask Colin Percival on Twitter, he's the person responsible.

Oh, mlu already said that. Sorry!

I will spend some time studying source code of blacklistd

blocklistd is somewhat different.

mage correctly pointed out that kqueue is the "BSD" way of doing things.

blocklistd is specifically made to have a daemon signal when a connection is attempted, so that blocklistd can be used to decide the rate at which too many connections are made, in an attempt to stop brute forcing, without relying on fail2ban or similar systems that require grep'ing log files

grep'ing log files for that sort of thing is a very inelegant solution, and requires considerably more cputime

yeah

I need to setup some user limiting to samba share. The only thing I could think of is set samba to log every file opening and deny access when exceeded limit

alright, so on a fbsd system, every time I lose power to the system, it goes to single user mode due to inconsistency in the fs.. this is undesirable. is there a way to prevent this? until I replace the battery in my ups this is going to be an ongoing issue.. not so much as I can replace it, but for future reference..

pvalenta: have you looked at rctl(4) and rctl(8)?

debdrup, I know about it but I need to study it deeper, if it can be helpful, thank you

illuminated, freedom from that is one thing ZFS gives you and UFS doesn't: on-the-fly repairs, with more extensive checks in the background every... I think it's 35 days by default.

I lost count of the times I had to hard-power-off my previous laptop when its keyboard and trackpad seized up, and never had any problem.

hrmm.. ok

I should have done a zfs on root

illuminated: its possible to migrate a ufs to a zfs without too much trouble

is there something like pciconf, that shows a hierarchical view of devices, as FreeBSD enumerates them?

so its clearly indented that Device X hangs off Bus Y

aah devinfo(8) thanks Rubber Duck Debuggers

dch: devinfo(8) is pretty good

Title: Net: add BSD ifconfig(8) parser and state class by igalic · Pull Request #1779 · canonical/cloud-init · GitHub

dch: also, the pciutils package contains lspci, which can display a tree

ta. I need base system tools for this one.

got it

devinfo or pciconf would be really neat with a ucl or json output

dch: You could see about fitting in libxo

dch: time to earn that src commit bit

I have a permanent tmux session on my machine, aber about 2 months of uptime the SSH connections to that tmux session become somewhat unresponsive, lagging for like 20seconds at a time sometimes

is that a known thing?

yourfate: two months of permanent connection?

no, on and off connections, but the tmux keeps running

yourfate: i have tmux sessions open for months or year+, never happens to me. so should be another issue

hmm ok

sad

is there a good way to profile this? htop etc also hangs when it hangs, so I don't see what happens

dunno, maybe someone will have some insight. do you have any hanging NFS mounts?

can you open new tmux panes/windows without the lag ?

oooh, I think its the weechat session

that gets laggy

its in a different tmux session

OK

and makes all other tmux sessions laggy somehow?

jmnbtslsQE: i wonder if you can do some kind memory monitoring to see if it's leaking

meena: at least you could use e.g. top to see about how much it uses after starting everything up, and compare that to a couple of days/interactions later :-)

jmnbtslsQE: maybe a massive scrollback buffer?

I stopped weechat, and it was fixed

I think it was the like 600 000 messages in there :p

:-}

yourfate buffer was filled?

well, was pretty full, but that should not make it hang

seems like I need to look into weechat config

I'm strange and have been using a combo of debian and smartos(illumos) for the last decade, but am learning/using freebsd. Using virtio nics for freebsd in a bhyve vm on illumos is slow, but e1000 is pretty fast. Is that expected behavior?

copec: there's bugs in the offloading code, if you disable that, it should be faster

illuminated: you could also set fsck_y_enable="YES" to force fsck to finish the job.

i usually use that and background_fsck="NO" for the few UFS root hosts i have

what does the latter do?

and does this get added to rc.conf or loader.conf?

probably rc would be my guess

rc.conf, yes

latter you will have to wait for fsck to finish before the system finishes booting

illuminated: sysrc -d mystery_variable_somebody_suggested

Title: sysrc(8)

Hello, I planning to install FreeBSD on a supermicro motherboard:

Title: A2SDi-16C-HLN4F | Motherboards | Products | Supermicro

And I'd like to use a PCIe x4 to M.2 adapter to boot from a NVMe drive in the adapter.

andr01d: I am proud of your decision

I'm naive on using M.2 drives, has anyone boot FreeBSD in this sort of way?

Does the M.2 drive show up in gpart or fdisk?

andr01d: I have an NVMe drive on my bsd machine but I don't know what can go different in your configuration. What do you worry?

andr01d: let me go check

I'd like to confirm that what I have in mind can generally work, before buying the motherboard.

mictty, Do the NVMe drives you use plug into M.2 slots?

andr01d: yes plug and no middle adaptor

My understanding is that the M.@ is just on the PCIe bus, and it "should" be the same using the adapter to a PCIe x4 slot...

Just wondering if anyone has done such a thing?

andr01d: in my case(without adaptor) gpart shows as 'nvd0'

thanks mictty ! I bet it's the same through the adapter. I'm going to order it and test it out.. Thanks Again!

andr01d: you can identify how it is recogized by the system via pciconf rather than dev abstraction

andr01d: my interface is found through 'nvme0@pci0' in pci conf.

andr01d: beside this, I don't know what I can help. I cannot test your configuration. good luck.

Is there a way to force a port to not build a dep? I have python3.10 installed, but postgresql14-server wants to bring in python3.9. Any takers?