Linux - GeneralThis Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
I have a question about configuring NTP in my network.
My setup looks like this:
Code:
---------------
| Main Server |
---------------
|
| over satellite
|
----------------
----------| central NTP |--------
| ---------------- |
| over satellite | over satellite
| |
| |
------------- -------------
| remote A | | remote B |
------------- -------------
| |
| local | local
| |
------------- -------------
| client A | | client B |
------------- -------------
I'm wondering if I should:
1. Have all remotes and clients and central, get their NTP from the Main server.
2. Have all clients get their NTP from the remotes, then have the remotes get thier NTP from the central, then have central get their ntp from the Main.
I have attempted to do option 2, but I'm having trouble getting the remotes to sync with the central. I can't get the remotes to sync with the central. This is a closed network so I'm not too worries about security.
The satellite links over central to remote are 2mb while from central to main is 8mb. I'm wondering also if doing option 1, would it eat up a lot of bandwidth. There is approximately 21 remotes, each with 15 clients, and one central and one Main.
well if you can get a "remote" to sync, why would a "client" sync??
Ideally an NTP config should reference at least 3 servers, so I would put all servers in all clients configs. Bandiwdth usage for NTP is neglibigle. Once fully synced up they would, by default, make a request (one very small UDP packet each way) to each server every 1024 seconds, which even with a lot of clients is very very little traffic.
to explore further why things aren't syncing, you should check output of debug commands like "ntpq -pn" on a running server.
Ya, the clients are not sync'd either, which would be another concern I have for single points of failure. So if there is neglible bandwidth usage by making all servers get it's timing from the Main then I can configure it that way. My only concern is that when the satellite links go down, which they do every now and then, is what I should do about the timing. Maybe I should add a local fudge server to their configuration?
I think the purpose of option 2 was for when the satellite links go down.
And all the other servers can access the Main but nothing beyond it, which the Main gets it's timing from three external sources.
For the central, the output of ntpq -p shows a (*) by the Main's IP, and I have a local fudge which is (space) by that.
If it's sync'd, would there be a (*) sign? My Main has 2 (+) signs and 1 (*) sign.
In all, for symbols on ntpq -p I have:
Main: 2+ and 1*
central: 1*
remote: 1*
clients: nothing
with the remote it's pointing to the Main and central. But it's got the (*) on Main, i think because it's a stratum 3 and the central is stratum 4.
I don't know why the clients don't have a (*) by the remote IP. It's the only NPT it's pointing to.
no, every client should have multiple servers, so when a link went down they would still be using their own local server by itself. no fudging. it's evil and always avoidable.
I have a closed network. the main server is the only one that has access to the internet where it gets 3 different timing sources. how would I configure the rest of the clients for 3 sources when I have only 1 reliable one?
well, you just add all three servers into your config. The most likely scenario is that all clients will sync back to the central one, but will still be able to happily switch over to a local one should the wan links go AWOL. That local server would want to be configured as an orphan to keep it happy if isolated... http://support.ntp.org/bin/view/Support/OrphanMode
Will the clients sync up on just one time source? Can you explain the (*) and the (+) symbol? I thought that the (*) meant it was a candidate but not sync'd, am I incorrect on that?
yeah + is the one it's using, * are other acceptable candidates which can be immediately used should + go awol.
Why won't orphan mode work? That comment about 4.2.4p5 says that if the config file lists NO servers at all then it won't load. You'll define three at least, so it'll be fine.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.