| Re: [chrony-dev] IPv6 link-local support |
[ Thread Index |
Date Index
| More chrony.tuxfamily.org/chrony-dev Archives
]
- To: chrony-dev@xxxxxxxxxxxxxxxxxxxx
- Subject: Re: [chrony-dev] IPv6 link-local support
- From: Miroslav Lichvar <mlichvar@xxxxxxxxxx>
- Date: Thu, 25 Jun 2020 15:59:37 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1593093583; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=j+0hAEsa4AwnSAqKdRZ6Cnh6STx/3V4r7LOipBxp1d8=; b=ecr6gFE4kAg4cFeCdOHgGuw7rXchGux4bnLmujVzGweKLi2Bgdx1Anb7/fDCdvg02aMkJm vtcfUnkWVhD8aajYQv/iLT6y+If518rxngsEiCoVUyin2YeV1ZRFU+/vx3/pHKypkqXeXY SErQuddZ2lIBiSOUPPHUVwCJuvRnXdw=
On Thu, Jun 25, 2020 at 01:10:35PM +0000, Scott, Nathan wrote:
> Hi folks,
>
> I work on the AWS Time Sync (https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/set-time.html) service. We create a link-local IPv4 address (169.254.169.123) that can be used as a source server for NTP daemons on EC2 instances. The future is IPv6, so we’re evaluating all of the clients and our current options. We will be using IPv6 link-local addresses.
As a side question, have you considered providing a virtual PTP clock
for guest synchronization, which could be used by chrony using the PHC
refclock driver? The Linux kernel supports this on KVM and Hyper-V.
> What is the work required to Chrony to support IPv6 link-local so customers can continue using Chrony instead of switching to an alternative? What are the next steps?
It depends on what exactly needs to be supported. A limited server
support is easy. A full server and client support is difficult as
there are many places that work with IP address alone (e.g. address
filter, client monitor, cmdmon protocol, chronyc) and which would need
to include the interface index or interface name. There are also
portability issues. ntpd has (a partial) support for link-local
addresses because it was designed to have a separate socket for each
interface. That is an anti-goal of chrony.
If it was acceptable to limit the support to a single interface, there
could be a bindacqiface directive to bind the client sockets to a
specific interface. Would this be sufficient for your users?
If more interfaces need to be supported, there could possibly be a
source specific option to bind the socket, but this alone would not
prevent a conflict between two sources that have the same address (in
different networks).
--
Miroslav Lichvar
--
To unsubscribe email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "unsubscribe" in the subject.
For help email chrony-dev-request@xxxxxxxxxxxxxxxxxxxx with "help" in the subject.
Trouble? Email listmaster@xxxxxxxxxxxxxxxxxxxx.