CentralConfig stops fetching entirely when there's a network error

[estolfo]

If a network error is encountered when fetching the remote config (i.e. a HTTP::ConnectionError), a bug in CentralConfig#schedule_next_fetch will skip scheduling the next fetch. Fetches are always scheduled by the last fetch in a chain, so as soon as a fetch has not been scheduled, the chain is broken and no future fetches will occur.

The culprit is in this line. When a network error is encountered, headers will be nil. An error would then be raised and the new @scheduled_task here will never be created.

Looking at the big picture, it might be safer and more efficient to use a TimerTask so a new ScheduledTask object is not created each time a config needs to be fetched and so that the "chain" dependency is avoided.\

Agent version is 3.6.0

[gaurav6192]

@estolfo is it resolved? In which version?

[estolfo]

Hi @gaurav6192 This is resolved as of agent version 3.7.0. Are you observing an issue with central config?