I've been contributing to open source since 1998.
| FreshPorts | https://freshports.org |
| homepage | https://langille.org |
| BSDCan | https://bsdcan.org |
| PGCon | https://pgcon.org |
I've been contributing to open source since 1998.
| FreshPorts | https://freshports.org |
| homepage | https://langille.org |
| BSDCan | https://bsdcan.org |
| PGCon | https://pgcon.org |
Two versions.
Dark mode: easy for me.
Light mode, nope, can't see it.
My database backups are all dumped locally, then copied, via a pull operation, to a centralized host (known as dbclone, a jail on a #FreeBSD host). From there, they are backed up to disk.
In addition, there is a script which runs through each of the backups and loads them into a database server.
Backups are grouped by host. For each host, the existing databases are cleared and started fresh with an entirely empty set of database files.
Here's the disk-io graph (the backups are still running). I was reviewing logs after some recent changes and noticed some problems. The spikes between 03:00 and 10:00 UTC are the cronjob.
At about 11:13, I restarted the cronjob after making some corrections. It takes about 8 hours to do all the databases.
This host has HDD. I'm tempted to migrate this jail from the existing host (r730-03) to another (r730-01) which is all sold-state drives.
see https://dan.langille.org/2024/07/16/wheres-my-backup/ for a bit more information on dbclone
This post shows more about the host: https://dan.langille.org/2025/06/30/r730-03-6/
Got #FreeBSD? Got #jails on FreeBSD?
Ever wonder how much resources a given jail consumes?
#LibreNMS has something for that: OS Level Virtualization
It's a FreeBSD port at net-mgmt/p5-OSLV-Monitor
Look at the following, which is testing database dumps to make sure they load. It is titled: CPU/System/User Time in secs/sec
Post incident, here's the #FreshPorts bump at midnight (UTC) mentioned in a recent previous post. It may not seem like much...
Looking back, it seems 0000 and 1200 are favored times.
More on #FreshPorts vs restrictions on search, etc.
Last night, I received a CPU alert regarding the database server.
I logged in and restricted search to logged-in users.
[13:15 aws-1 dvl ~] % sudo grep 'joe' /var/log/auth.log.1
May 4 00:57:46 aws-1 sudo[37761]: dvl : TTY=pts/0 ; PWD=/usr/home/dvl ; USER=root ; COMMAND=/usr/local/bin/joe /jails/nginx01/usr/local/etc/freshports/constants.local.php
This change means anyone not logged into FreshPorts with their FreshPorts account will get a 401 for any search.
This morning, the graph for Nginx requests graph looks like this.
Search was enabled again at May 4 13:14:37