Bug 197393
| Summary: | Default installation should have provision for correct time | ||
|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | David Anderson <david> |
| Component: | ntp | Assignee: | Miroslav Lichvar <mlichvar> |
| Status: | CLOSED WONTFIX | QA Contact: | Brian Brock <bbrock> |
| Severity: | medium | Docs Contact: | |
| Priority: | medium | ||
| Version: | 5 | ||
| Target Milestone: | --- | ||
| Target Release: | --- | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: | ||
| Last Closed: | 2006-07-26 14:48:23 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
I agree it would be convenient to many users to have something like this enabled by default, but it's unlikely to happen. Running ntpdate from cron is generally a bad idea. It will not keep very good time at first place, after every ntpdate call the clock will drift again. However, the reason why ntpdate will never be in crontab by default is that it would cause a major network problem, imagine that all the systems connected to internet in a timezone query a server in the same time. Running ntpd is a better idea, but I still think it is better to have it disabled by default. Drifting: That's not a huge problem. We're talking here about users who don't have the expertise or don't care enough about the precision enough to run ntpd. We're talking about users who just want their clocks not to run into minutes of drift. I run ntpdate hourly via cron. Windows XP seems to sync via SNTP weekly. It's "good enough". Anyone running a server should run ntpd, I agree. My machines, with ntpdate hourly, drift by about 0.4 seconds an hour. If I cared about that I'd run ntpd, I'm not running a server, so it's fine. Major network problem: Fedora already faces this issue with automated yum updates (and ClamAV database updates in Extras). In both cases, it introduces a randomised delay in order to avoid the problem. We could do the same here. Please consider re-opening! The SNTP implemented in Windows XP is a service running all the time. Please drop the idea about putting ntpdate into the crontab. Enabling the ntpd service is easy. Even user who doesn't know anything about NTP will notice the "Network Time Protocol" tab in the system-config-date dialog. You can file a bug for system-config-date if you think it's not easy or obvious enough. BTW, there is a SNTP utility in ntp-4.2.2 packages, unfortunately it's not ready to be a service yet. |
IMO, it should not be considered acceptable nowadays that after installing Fedora Core, manual action is needed in order to get correct time. Windows XP has had automatic SNTP since 2001. I suggest a cron.hourly or cron.daily cron job that syncs time. This is better than running the NTP daemon permanently, because we do not want to add extra services to run by default on startup. The only problem with this would be if the user sets up NTP manually, he may not realise he needs to turn off the manual job. But this would be easy to deal with - the cron script can check if NTP daemon is running or not. Suggested cron.hourly (or daily) script: /sbin/service ntpd status >/dev/null || /usr/sbin/ntpdate -s clock.redhat.com >/dev/null 2>/dev/null Alternatively, the script could check if NTP is _meant_ to be running (by looking for the PID file), to deal with the case that NTP has crashed (the admin may not wish ntpdate to be running in this case). Maybe you'd want a variable and a file in /etc/sysconfig/ in order to turn this off, but I cannot think of any reason why anyone would want to turn it off. The only reason to turn it off would be because the user is running ntpd.