The first steps to avoid this warning:

poudriere: Warning: Using ‘-‘ in a SETNAME is not recommended as it causes ambiguities with parsing the build name of 122amd64-default-master-list

Today I've been using 'mkjail update -a' to update all my jails with respect to recent published known vulns on .

Why update the jail and the host? Because I can.

Patch everything when a patch is available. One vuln might have unknown leverage. I can't tell. So I patch.

How is it that I've been writing shell scripts for over 15 years now to automate processes, and I've only just discovered $SUDO_USER today? I feel like that's a MASSIVE oversight.

Is this a new device or not?

gpart shows a partition.

smarctl says 4 power on hours, 56 power cycles, 104GB written.

Could that just be initial testing at the factory?

I configured an Nginx proxy between 1 host & the iDRAC of a host in another datacenter. The ISP has the iDRAC IP access filtered to a small set of IP addresses.

Proxy is available only via VPN.

I have only one host at the 2nd datacenter. No other control.

This situation limits the available solutions & not all those mentioned in the original thread apply. This solution is self-hosted and seems to work well.

Looking for suggestions/improvements.

Details here:

Today, while implementing a new 'when were the reports last sent?' monitoring check for @FreshPorts, I discovered that the 'Ports I maintain' report last went out on 2020-08-20.


This report lets you know if any commits occur to ports you maintain.

It should go out tonight.

I composed a multi-part tweet earlier today and it disappeared when I dropped an image into the browser instead of the composition window.

Instead, you must now suffer through a blog post.

With graphs. And scripts.


Done. My thanks to Alan Hicks for helping with this.

The css is:

.commit-branch {
border-style: solid;
border-radius: 6px;
padding: 0 6px;
background-color: darkred;
border-style: none;
color: white;

Show thread

CSS help wanted for @FreshPorts.

There are 2x 11:16 entries at the bottom of this screen shot. Beside them, you see 2021Q3 in a border.

I want that text (2021Q3) very obvious to indicate that this commit was on a quarterly branch.

Can you help with that please? Thank you.

Tonight, for the first time in 29 days, one of my servers is free of known vulns.

What changed?

I removed the mysql56 jail. MySQL 5.6 went EOL in February 2021. I've been running Bacula regression scripts against it.

But no more. Less red. More green. On Nagios.

During an online meeting, discussion about recordsize for zroot/var/log on came up. Why not 1M instead of 128KB?

Some tests were done. Looked good.

Tonight I did this on all hosts:

$ sudo zfs set recordsize 1m zroot/var/log

Spend a few minutes today to save yourself tomorrow.

It's not so much the odd errors, but the reasons for doing something a particular way are vital to understanding the choices. When you, or someone else, looks at the system years later, they wonder: why was that option selected for the function call?

A comment solves that.


Show thread

I remember asking a team member, in the early 1990s:

me: maybe we should document the odd problems we encounter and how we solved them?

them: Chances are we'd never encounter that problem again, so what's it worth?

Don't underestimate the investment for future you.

Many times I have hit a problem and not known how to solve it. Searching the internet, I find an old blog post of mine. Not only have I hit the problem again, I've completely forgotten about solving it previously.


The latest addition to my monitoring.

Are all installed packages available for reinstall?

orphans=$(pkg version -vRL= | grep orphaned)
if [ -z "$orphans" ]
echo "All packages are in the upstream repo"
exit 0
echo $orphans
exit 2

PGCON is 11 days away on June 28

For 2021, the event is reduced to the Unconference only. There will be no formal talks this year. Looking ahead to 2022, we see the conference returning to the usual format.

Details coming later via Twitter & webpage.

Anyone looking to hire?

* IT Professional
* Unix/LINUX RedHat RHEL Networking
* Data Center
* Deployments
* Escalations
* Customer Relations
* Data Management
* Troubleshooting


I want a list of all the children of a filesystem. I'm using:

$ zfs list -Hr -o name -d 1 foo/bar

I don't want the parent in that list (i.e. foo/bar). Can `zfs list` do that?

If not, I can use grep to get rid of `foo/bar`.

The output I want is:

To initiate processing of git commits in the ports tree, the last remaining change is to this line in /usr/local/etc/


But first, I have this bagel in front of me.

I'll work on configuring FreshPorts for the new git ports tree at about 1600 UTC. Lunch time.

Show older

Fosstodon is an English speaking Mastodon instance that is open to anyone who is interested in technology; particularly free & open source software.