Cronicle reports "server clock is seconds out of sync"

[SprickW]

I'm running Cronicle server along other LXC/VMs in a Proxmox cluster.

Chronicle reports for several client systems "server clock is x seconds out of sync", x sometimes being > 10.

On Proxmox VM/LXC receive their time from Proxmox hosts and manually checking I could not find any different time at VM/LXC level and Proxmox server from real time.

Any explanation for the errors reported by Chronicle?

Kind regards Wolfgang

[jhuckaby]

Sounds like your clocks are definitely drifting. Cronicle just compares the raw absolute epoch timestamp of the worker servers vs. the primary server, and if they are 10+ seconds off from each other, it logs an error.

This isn't a fatal error, but rather just a warning.

I will make this configurable in the next version, so you can set it higher than 10 seconds to silence the errors. But this is definitely a sign of a problem in your environment, and warrants a closer look. If the clocks drift too far off from each other, jobs may not start or finish on time, etc.

Please check the NTP service (ntpd) on all your servers, and make sure they are all pulling from the same time source (e.g. time.apple.com). I understand you are running containers, and those SHOULD keep in sync with their hosts, so maybe the problem is on your host machines?

[jhuckaby]

In Cronicle v0.9.72 which was just released, you can now set a max_clock_drift configuration property in your config.json file (or set it with a CRONICLE_max_clock_drift environment variable). This will allow you to customize the maximum allowed clock drift -- the default is 10 (seconds).