kevans: nice. i have pretty minimal jail around 89MB in size with custom SRCCONF and SRC_ENV_CONF. it started and works. the pkgbase variant with FreeBSD-set-jail-minimal maybe little bit smaller but i not completely familiar with pkgbase.

I found something odd.. in FreeBSD, if you're running php_fpm, and you send it a Hangup signal (kill -1 / -HUP), it just kills php_fpm completely. With its debug messages on I see 'Hangup' in the terminal, followed by: /usr/local/etc/rc.d/php_fpm: WARNING: failed to start php_fpm. On Linux, sending signal 1 to the php "master" process just causes it to reload. Anyone know why? It's not a socket or network

issue (tried with both and it does the same thing on both)

damn.. I guess I need to use SIGUSR1 not HUP

pertho: I run php_fpm here and there, but I mostly manage it via the startup scripts. I guess you meant rotating logs with SIGHUP, and so I looked it up in one of my jails, I use SIGUSR1 for that

wow is it me or do passkeys fucking suck?

It's not you.

karolyi: yeah I just changed the newsyslog.conf entry from '1' to '30' (last field) .. which should do the trick

it's been years since I read about php-fpm and signals.. :D

USR2?

USR2 for reload

USR1 for log rotation

USR2 to reload config/children

yea

HUP just kills it dead :D

hooray for consistency! :D

seems a bit unintuitive for a program w/o a controlling terminal. if being run with a controlling terminal HUP killing makes sense, as HUP is the signal sent when the phone line clicks

Heh.. someone on Linux-side wrote a little shell script called /usr/lib/php/php8.x-fpm-reopenlogs .. which looks for the pidfile, and sends kill -USR1 to the number. newsyslog is a lot more capable than logrotate :D

which linux-side? debian?

ubuntu/debian

or, at least, in the Ondřej Surý packages

ah

the php_fpm rc script also has php_fpm_logrotate() so I guess you could just call /usr/local/etc/rc.d/php_fpm logrotate ?

just use SIGUSR1 from newsyslog

also, if you think php_fpm logrotating is a PITA, ask me about rotating uwsgi logs.

hm, now that I mentioned it, I must just make a MR for the project to support it

karolyi: yeah I just used this: /var/log/php-fpm.log root:wheel 644 14 * @T00 GBJC /var/run/php-fpm.pid 30

I should just consolidate into one line and make it /var/log/*php*.log

because I have a php-fpm access log as well

pertho: /home/website-wp/wordpress/log/*.log website-wp:website-wp 644 30 * $D5 JG /var/run/php-fpm.pid SIGUSR1

is what I have. good luck

i wonder if i should do some crimes to php_fpm and uwsgi to obviate the logrotation gluteache

alias add b /buffer

karolyi: thanks! :)

angry_vincent: nice!

if I wanted to clear VERASE2, how would I theoretically go about that

theoretically stty erase2 undef would clear it, disables the erase2 control char entirely by setting it to _POSIX_VDISABLE

hi folks, wondered if somebody can tell me if I have missed a step. I have configured a number of vlan interfaces on parent interface re2. There's no traffic passing into my switch from vlan 160 so to t-shoot I am looking for vlan tagged traffic on the freebsd box with `tcpdump -envi re vlan 160` however nothing is showing up. If I omit the traffic id of 160 I can see all the other vlan traffic belonging to different vlans.

How might I further discover what is going awry?

(typo: `tcpdump -envi re2 vlan 160`)

Been a minute since I did VLANs on FreeBSD.

I seem to remember though, the vlan interface should have its own "NIC" name that you would run tcpdump against.

correct--you end up with ix0.1234 as a vlan interface off of ix0. you can also tcpdump -i ix0 vlan 1234 to see the same traffic as tcpdump -i ix0.1234

skuzzee: pastebin your relavent rc.conf bits as well as the switch port config

I miss all the different things I used to do with FreeBSD when I was at EMC.

All sorts of crazy network appliances.

LACP link-aggregation, VLANs, a "WAN emulator", an in-line sniffer, an ntp server, that could be set to arbitrary times to test how clients would react to it.

Is it wise to use #!/usr/bin/env sh for easy bsd support or should by default I expect bsd developers to change the shebang

hdconfigure_: any reason not to just use #!/bin/sh?

I thought #!/bin/sh does not work on freebsd

works fine, but the key thing to be aware of is that /bin/sh is not bash

so, if your script uses bashisms, you would need #!/usr/bin/env bash

ohhhh that explains why I was confused a few years ago. is freebsd posix compliant?

/bin/sh is indeed POSIX compliant

so my language model is saying to use #!/usr/bin/env -S program --option value

does this seem like the best way if I want the shebang to run any program in "$PATH" with any arguments

if your script needs a POSIX-compliant shell, just use #!/bin/sh. if you use bashisms, use #!/usr/bin/env bash

If all else fails I can use #!/bin/sh and then launch my program from within sh but I want to avoid that redirection and use shebang if possible

my only concern is the portability of this form #!/usr/bin/env -S program --option value

what is program?

any executable that I can launch in sh by typing $/path/to/my/exceutable/executablename.filextension arg1 arg2 argn...

oook, ill leave you to it then

thanks you have helped me alot already. I am gonna move this question to #bash

apparently that form I posted with the -S option is a gnuism

well, freebsds env supports it too, but it isn't in posix, so check if all your target operating systems support it