Bug 1975415

Summary:	With no refclocks at all, ntpsec can't be fed from ntp-refclock or gpsd
Product:	[Fedora] Fedora	Reporter:	George Joseph <g.devel>
Component:	ntpsec	Assignee:	Miroslav Lichvar <mlichvar>
Status:	CLOSED DUPLICATE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	34	CC:	mlichvar
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-06-23 16:45:58 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description George Joseph 2021-06-23 15:51:03 UTC

Description of problem:

Since ntpsec is compiled without any refclocks, including SHM, it can't be fed by gpsd or ntp-refclock.  It also doesn't print any messages to indicate that refclock support is disabled.  

Version-Release number of selected component (if applicable):

1.2.0-6.fc34


How reproducible:

Continual

Steps to Reproduce:
1. Attach a GPS unit to the system.
2. Start gpsd with a path to the device.
3. Add "refclock shm unit 0 refid GPS" to /etc/ntp.conf
4. Start ntpd -d
5. Run ntpq -p


Actual results:

No error messages concerning the refclock and no peers recognized 

Expected results:

ntpq -p shows a peer of *SHM(0)

Additional info:

The implications of splitting the refclocks out of ntpd was never really explained.  Without refclocks, ntpd has become just a network forwarder which is a significant step down from it's former capabilities for those who need to do more than just that.  Combine this with the fact that ntp-refclock only supports  socket connectivity and not shared memory makes things even more difficult and complicated.  I'm not even sure what the point of ntp-refclock is given that gpsd already does the same thing, plus more.  Actually, I'm not sure what the point of a stripped down ntpsec is since users who just need a forwarder can use chronyd.

Comment 1 Miroslav Lichvar 2021-06-23 16:14:02 UTC

It wasn't intended for ntpsec to have the refclock support disabled. This was already reported in the bug #1955859 and it should be fixed in this update waiting to reach the stable threshold.

https://bodhi.fedoraproject.org/updates/FEDORA-2021-3ffc890685

Comment 2 George Joseph 2021-06-23 16:45:58 UTC

(In reply to Miroslav Lichvar from comment #1)
> It wasn't intended for ntpsec to have the refclock support disabled. This
> was already reported in the bug #1955859 and it should be fixed in this
> update waiting to reach the stable threshold.
> 
> https://bodhi.fedoraproject.org/updates/FEDORA-2021-3ffc890685

Ah, I saw that bug report but didn't associate the title "ntpsec: ntpkeygen generates weak keys" with not having refclock support. :)
It also explains why I saw '--refclock=all' in the spec file in the repo but not in the spec file in the current srpm.
I can test the new package and give it a vote.

*** This bug has been marked as a duplicate of bug 1955859 ***