Unexplained difference between left-side column timestamp (Nodes view) and tooltip timestamps

Hello,

Thanks for confirming the bug :D

PS : I think this question can be tagged as resolved but I have no idea how to do that on this forum

Hi ​@BugHunter42, there’s a “best answer” gray button below the text. I marked that off, so no worries.

See you around!

​@xamici Thanks

This is a classic Rundeck synchronization puzzle. The 14-hour gap suggests that while your -Duser.timezone is set, there might be a conflict with how the database (where Rundeck stores job metadata) and the local log storage handle UTC offsets. Often, the sidebar reflects the UI/Browser interpretation of the server time, whereas the .rdlog reflects the raw timestamp from the execution node.

I’ve encountered similar synchronization hurdles while debugging automated scripts for this page, and it usually boils down to the database connection string missing the serverTimezone property or an offset mismatch on the remote node itself. Have you checked if the remote node's hardware clock is drifting, even if NTP is active? Understanding this inconsistency is key to maintaining reliable execution logs.

Hello,

Thanks for the info. However, database timezone is set correctly as well.

In rundeck-config.properties file, the dataSource.url= URI does contain serverTimezone=Europe/Paris.

Obviously, there isn’t 14 hours offset on the hardware clock either, I would ahve noticed it.

Here’s the tests from this afternoon, on both the Rundeck server and Rundeck database at the same time (almost, minor network latency) with multissh.

Rundeck server :

[root@rundeck_server]# timedatectl                      
               Local time: lun. 2026-04-20 14:32:55 CEST
           Universal time: lun. 2026-04-20 12:32:55 UTC
                 RTC time: lun. 2026-04-20 12:32:55
                Time zone: Europe/Paris (CEST, +0200)
System clock synchronized: yes
              NTP service: active
          RTC in local TZ: no
[root@rundeck_server]# hwclock --show; date +"%Y-%m-%dT%H:%M:%S %Z"
2026-04-20 14:33:11.997876+02:00
2026-04-20T14:33:13 CEST

Rundeck database server:

[root@rundeck_server]# timedatectl                          
               Local time: Mon 2026-04-20 14:32:55 CEST
           Universal time: Mon 2026-04-20 12:32:55 UTC
                 RTC time: Mon 2026-04-20 12:32:55
                Time zone: Europe/Paris (CEST, +0200)
System clock synchronized: yes
              NTP service: active
          RTC in local TZ: no
[root@rundeck_server]# sudo hwclock --show; date +"%Y-%m-%dT%H:%M:%S %Z"
2026-04-20 14:33:12.997588+02:00
2026-04-20T14:33:14 CEST

Remote note from rundeck adhoc commands:

Command Execute a remote command
Command: timedatectl
#4719999
               Local time: Mon 2026-04-20 14:44:25 CEST
           Universal time: Mon 2026-04-20 12:44:25 UTC
                 RTC time: Mon 2026-04-20 12:44:26
                Time zone: Europe/Paris (CEST, +0200)
System clock synchronized: no
              NTP service: active
          RTC in local TZ: no

Since the remote node is not one of my servers (I don’t have a direct SSH access to it, outside of Rundeck) and the Runckuser is not a sudoer, I can only use timedatectl, not hwclock --show. So unless proven otherwise, I’m assuming hardware clock as seen by systemd’s timedatectl (RTC Time) is as good as hwclock --show on the remote node, just like is it on both Rundeck server and Rundeck’s database server.

PS : From hwclock --help, the --show option displays the hardware clock as-is, uncorrected, while --get option displays the hardware clock with drift corrected.

Functions:
 -r, --show                      display the RTC time
     --get                       display drift corrected RTC time