Increased data request latencies

This incident has been resolved. We recommend that Agents are upgraded to v2.6.2 on the stable channel, or the latest nightly build at the earliest convenience.

We've implement a change for this issue which appeared in our latest nightly, and resolved the problem. We are now issuing a new stable release v2.6.2 to include that fix, and will resolve this issue when all packages are available.

We have identified the cause of this issue and have been creating a fix which will appear in the upcoming nightly build. When we are confident that there are no negative effects on that build, we will do a new patch release on the stable channel the next day.

What happened is that we were not correctly clearing MQTT messages that had been acknowledged by the broker, causing newer messages to not be correctly recognized as acknowledged in certain conditions.

We will be adding additional debugging code in both the Agent (next nightly) and Cloud Backend to track what is causing these spurious latencies.

We are investigating an issue with increased data request latencies that can result in slow or completely empty charts for some customers. The affected Netdata Agent versions are v2.6.0 and later, for all stable and nightly builds.