A while back, I build my packages for each of Python 3.12, 3.13, and 3.14 to see which version I would target as I moved off Python 3.11

I chose 3.12 as it build everything I needed.

I noticed today that my daily builds against those three versions continue. I'm going to stop my test build against 3.12 (because I'm now using it daily). The builds against 3.13 and 3.14 will help me keep track of what's building and not building.

That information will let me know when it's time to move to a newer version of Python.

@mason

Thanks, fixed it.

I just updated a FreeBSD 14.3 jail to FreeBSD 15.0 using mkjail. This is unformatted. Scroll to the bottom for the actual update.

In short, use the FreeBSD pkg repo until you get your host/poudriere running.

https://dan.langille.org/2026/02/08/updating-pkg01-to-build-15-0-packages/

Not shown, the jail host was updated first.

I'm on #FreeBSD 14.3 - if I move to 15, it will be on my main dev box: r730-01 - that is always the first host to get updated. Then my package build jail, pkg01 - so I can build my 15.0 packages for that host.

To be clear, I mean the #FreeBSD packages built by my poudriere jail, pkg01.

When I previously went through this (in my mind, but writing it down here), a reply claimed that I should be able to build 15.0 packages now for my upcoming upgrade to 15.

I'm familiar with building FreeBSD itself (i.e. I known I can build FreeBSD 15.0 on my FreeBSD 14.3 host). That is rather distinct from building the packages. I still think I have to be booted and running FreeBSD 15.0 in order to create 15.0 packages.

I shouldn't be so concerned about this. The pkg01 jail started off on FreeBSD 12.0 as I just searched and found here: https://dan.langille.org/2019/10/23/moving-poudriere-from-the-host-into-a-jail/

That system has had at least two hardware upgrades along with name changes. You can find pkg01 on each of these posts:

0 - Dell R710 - see URL above and https://dan.langille.org/2019/11/08/r720-3/
1 - Dell R720 - https://dan.langille.org/2019/11/08/r720-3/
2 - Dell R730 - (current location) https://dan.langille.org/2025/12/10/r730-01-3/

Over that same time period, the host has had two major OS upgrades:

From 12 -> 13 -> 14

Each time I've managed to get the packages done.

I don't know why I'm hesitant this time. I think it is mostly pkgbase.

I want to start building my own #pkgbase... mostly because I have my own poudriere server.

Perhaps I should not be so hesitant and just not build pkgbase and use the FreeBSD provided pkgbase.

Enough of that thought process. Time for a walk in this sunny chilly weather.

Thanks for coming to my TED talk.

@jmock I may have known, but forgot. I see Portland mentioned often... I recently thought you lived out west. Or perhaps West Virginia....

@jmock Today I learn we're in the same state.

@mwl I’m sad for them.

zroot is the basis of Boot Environments.

BE allows upgrade in place with greatly reduced risk; there is a known good fallback position.

Hopefully, one day they can get their shit together.

Wow, getting Chrome for Linux on #FreeBSD is a breeze now.

1. linux_enable="YES" in rc.conf
2. pkg install linux-chrome
3. Start Chrome

Bing Bang boom

Today I removed devel/freebsd-git-devtools from a buildlist I have in #poudriere - it has been renamed.

I removed it because it wasn't installed on any of my hosts.

I conclude that by this query:

samdrucker=# select host from hostswithpackageshowversion('freebsd-git-devtools');
host
------
(0 rows)

Then I got to thinking: where was it installed and when?

The comment in the buildlist said:

-# for working on FreeBSD
-
-devel/freebsd-git-devtools

I went back to the SamDrucker database and struggled parsing the JSON for a while. Eventually I came up with this inefficient but effective query:

samdrucker=# select date_added, client_ip from incoming_packages where data::text like '%freebsd-git-devtools%' order by date_added desc;
date_added | client_ip
----------------------------+---------------
2024-01-17 03:05:26.208061 | 10.55.0.29/32
2024-01-16 03:24:16.145212 | 10.55.0.29/32
2024-01-15 03:39:19.388747 | 10.55.0.29/32
2024-01-14 03:58:35.967538 | 10.55.0.29/32
...
2023-04-01 03:35:54.836752 | 10.55.0.29/32
2023-03-31 03:59:51.397951 | 10.55.0.29/32
2023-03-30 03:52:11.768713 | 10.55.0.29/32

That tells me it was installed on one host for about 10 months. That host is pkg01, my package build server.

This type of search was one of the orignal goals for SamDrucker, but having historical data has helped me here.

SamDrucker was written for FreeBSD, but it wouldn't take much to port. Both client and server are lightweight and easily understood.

The PostgreSQL function might take a bit of work if you want to migrate away from that choice. The reset is all shell/lua.

https://github.com/dlangille/SamDrucker

@fwaggle I'm not convinced that works.

If it does not, perhaps try a firmware update. I read about that too.

@stefano Is that imposter syndrome?

A note from our editors:

@feld And we have a rabbit hole:

https://forums.freebsd.org/threads/parallel-upgrading-of-jails-results-in-proposed-deletion-of-etc-passwd.91260/

@feld I realized my gateway is the only host at home without jails.

Let's test it there so I don't mess anything up.

I've created two jails over there.

However, I can't test the update right now because they're both fully updated.

[20:56 gw01 dvl ~] % zfs get -r mkjail:version zroot/jails
NAME PROPERTY VALUE SOURCE
zroot/jails mkjail:version - -
zroot/jails/ex1 mkjail:version 14.3-RELEASE-p8 local
zroot/jails/ex2 mkjail:version 14.3-RELEASE-p8 local

Damn you mkjail!

I hope to remember this after the next security announcement.

@feld No, I was waiting for you to do that.

I suspect it does break because each one will clobber the others diff/etc.

@mWare Without knowing, I'd say you dump the database and move it over.

I've never done that.

Here's a LibreNMS tip. It's really an snmpd tip

If you have three hosts, and one of them is not acting like the others, check the snmpd settings. In my case, the third host was configured with ucp. Swapping to tcp meant the bind application starting showing up.

@feld Imagine running 35 in parallel....

@feld have you tried running mkjail in parallel?

How many users do I mean by "a few?" When synchronizing UIDs across all of your systems begins to really, really annoy you, you no longer have a few users.