Red Hat Bugzilla – Bug 68039
firewall makes ntp setup useless
Last modified: 2008-05-01 11:38:02 EDT
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.1) Gecko/20020625
Description of problem:
Firstboot gives the user a possibility to configure the ntp service, but the
standard firewall blocks ntp in Medium or High levels if the user doesn't know
to customize it to allow udp:ntp.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
2.At first boot when the configuration utility runs, set firewall to Medium or
High, and and choose a ntp server.
Actual Results: ntp can't connect to the time server
Expected Results: configuration utility should open firewall for the selected
ntp server or connection tracking should be used so ntp would work.
This screen is actually dateconfig. The UI is just pulled into the firstboot
Here's the problem. We don't have a real firewall API at the moment, so it's
hard to tell exactly what the firewall is set to, and I'm reluctant to have
dateconfig poke a hole in the firewall without telling the user what is happening.
I've made some recent changes to dateconfig that check to see if we can ping the
ntp server before starting ntpd. If the server cannot be contacted, then a
dialog appears telling the user the either the specified server isn't available
or the firewall is blocking ntp connections.
This isn't a perfect situation, but it is at least an improvement.
I added a warning in the docs telling how to use redhat-config-securitylevel to
allow connections through the NTP port.
QA, please verify that redhat-config-date-1.4-7 informs the user if the server
cannot be contacted. As I said earlier, I don't think that it's the correct
thing for redhat-config-date to modify the user's firewall settings, so I'm not
going to implement that at this point.
With redhat-config-date-1.4-7, I'm getting a failure to sync with the timeserver
in the case detailed above. I'm not getting any type of message that the config
tool is unable to contact the timeserver.
Can you demonstrate this behavior for me?
OK, what I'm seeing is that if you provide a valid host (time.nist.gov for
example) but your firewall rules are preventing traffic on port 123 (ntpd) then
the sync is failing, but the user is never notified that the sync failed.
Should be fixed in 1.5.2-5. Please verify.
Here's what I'm seeing with redhat-config-date-1.5.2-5. When running firstboot
with the 'high'security level and entering an invalid host, I'm getting a
message informing me of that. If, when running firstboot, I enter a valid host,
but am unable to connect to it as a result of the iptables rules, I am not
getting a message informing me of that. The sync just fails.
Now, if I run redhat-config-date stand-alone, then I don't get a message either
way. So, still looks like some things need to be cleared up.
I think you need to update your ntp. The latest ntp package should use the
initscript to poke a hole in the firewall, so the firewall settings should never
be able to keep the ntp sync from succeeding.
Fix confirmed with ntp-4.1.1a-7.