1

I have four VMs that are all having CPU usage spikes every 10 minutes. These spikes coincide perfectly with DNS lookups for hostnames including archive.canonical.com and security.ubunutu.com (per tcpdump), so I assume apt is checking for updates. It's configured to check for updates daily. I can't figure out what's causing this 10-minute interval.

I've looked in /var/log/unattended-upgrades/, /var/log/dpkg.log, /var/log/apt and journalctl and can't find anything corresponding to a 10-minute interval.

These VMs are stock Google Cloud Ubuntu 19.10 images with very few changes (MongoDB installed [which added repo.mongodb.org as an apt source], logging and monitoring agents installed, THP disabled).

What should I do/where should I look to troubleshoot this?

3
  • 1
    Start with ls -l /var/lib/apt/periodic. That will tell you the last time unattended-upgrades ran. Next, try systemctl list-timers apt-daily.timer apt-daily-upgrade.timer for the next time apt will update/upgrade automatically.
    – user535733
    Apr 7, 2020 at 21:37
  • @user535733 Thanks. /var/lib/apt/periodic/update-success-stamp's timestamp coincides with the 10-min interval. unattended-upgrades-stamp and all others are much older. Those two systemd timers have intervals of 6h and 24h. None of the other systemd timers have short intervals either. -- So the question remains, why is apt updating ever 10 mins...
    – ZachB
    Apr 7, 2020 at 22:21
  • Oddly ... there's an entry in /var/log/apt/history.log from yesterday for Commandline: /usr/bin/apt-get remove -y unattended-upgrades.
    – ZachB
    Apr 7, 2020 at 22:30

1 Answer 1

1

Urg, this is the Google osconfig-agent. Apparently it polls every 10 minutes and disables automatic apt updates (re: my comment about the surprising apt-get remove -y unattended-upgrades line in the apt history).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .