NTP1.usv.ro - OPERATING STATUS / DEVELOPMENT INFO / ACCESS RESTRICTIONS

(25/01/2013) - Importance:

A catastrophic failure of the internal HDD occurred on the 12th of September 2012.
The server has been unavailable until today. We replaced the HDD with a Maxtor 31024H1 10GB (9641 MB to be more precise) HDD (ATA / PIO 4),
similar to the damaged one. We reinstalled the FreeBSD 5.4 operating system, recompiled the kernel with PPS support, and configured all the services.
All users having ntp1.usv.ro in their ntpd.conf are kindly advised to restart the ntpd service on their machines.

(12/07/2012) - Importance:

All time servers are able to serve both IPv4 and IPv6 requests.

(18/06/2011) - Importance:

We made several tests with IPv6  protocol. All things are in order with the exception of a switch which is not IPv6 compatible (I said "switch").
So, if you plan to implement IPv6, the AT-GS950/24 from Allied Telesis is a bad choice. After we will change the switch, our time servers will be
available also on IPv6, as all configuration changes were made and tested with success.

(15/06/2011) - Importance:

The IP address of the server changed to 80.96.120.251. For good results always use the DNS name and not the IP address. If you manually put
ntp1.usv.ro in the ntpd.conf file, you need to restart the ntpd service in order for the new IP to be used.

(06/06/2011) - Importance:

We've updated the firmware for the GPS receivers, from version 2.80 to the latest 3.30. No service interruptions occured as the servers were in sync
with the Rubidium PPS source.

(03/06/2011) - Importance:

We changed the servers location in a new building and positioned the GPS receivers on the rooftop, with almost 100% visibility to the sky. This means
improved performance compared to the old instalation. The hardware configuration didn't changed.

For the next 4 months, Internet connection failures may occur, so service interruptions are inevitable. Even our Internet connection will be down,
the time servers will work in the background without degraded performances, as we use no Internet information to keep the time accurate.

(23/05/2011) - Importance:

The servers location will change starting the 1st of June 2011. Beginning the 23rd of May 2011, we planned a series of tests in order to find the best
configuration for the servers for the period with limited or no GPS receivers visibility to the sky. The PPS Rubidium source will be available as a sync
source. For the next 4 months, Internet connection failures may occur, so service interruptions are inevitable.

(25/12/2007) - Importance:

I think there is enough time I ran the time servers to conclude that with some money and a lot of passion one could set up a very good time server.
A simple "top" command could show us some interesting data, for example the uptime of the server. Here is the screen capture (yes, there are more than
583 days of uninterrupted running time). Too bad for Netcraft that was unable to spot the server uptime correctly.

P.S. I didn't modified the screenshot (you may see the usernames). I am quite sure with the settings in hosts.allow and ipfw.conf there are very small
chances for bad guys the break the server (the main idea is that this is a server providing a public service - free of charge).

(19/12/2007) - Importance:

We changed the main ISP link from AS6830 to AS2614 (via AS20965) having now a very good international connection (with several external active
routes and peering). The maximum download / upload speed is now up to 30 Mbps, more than the previous limit. By the end of this year we have as
a target to upgrade the national connection between Suceava and the Roedu.net backbone to 1 Gbps. The total upload / download limit for the
external link (GEANT2 connected) of the central network backbone in Bucharest is 2 x 622 Mbps.

Post event information: The server was unavailable for various time intervals, due to route propagation process, between 06:25UTC to 19:00UTC.

(22/02/2007) - Importance:

I have installed one of our PRS-10 Rubidium Frequency Standard between the PPS output of the GPS receiver and the PPS input of the server.
Nothing changed in the server behavior. I made some test by interrupting the PPS signal from the GPS receiver and, as expected, nothing changed, as
the PRS-10 was synchronized and in phase with the PPS signal. The interruption took 6 hours and there was no signs in the log files and ntpd
parameters the server was affected. This is the first phase of the test; in the second phase I plan to replace the crystal on the motherboard with
a ICS-525 programmable synthesizer.
During this operation the server availability was not affected.

(30/10/2006) - Importance:

We modified the ntp.conf file in order to allow the server to synchronize ONLY from the GPS receiver
(and the associated PPS signal). As a result, when the GPS signal will not be available, the server will
respond with a "NOSYNC" message to the NTP requests. This is an intermediate phase before configuring our new
Rubidium PRS-10 frequency standards (when the error will be in the Stratum 1 interval for 72 hours in the absence
of the GPS signal). We also intend to modify the motherboards of our time servers in order to replace the 14.318182 MHz
crystals with a ICS525 Programmable clock synthesizer.

(27/09/2006) - Importance:

During the time interval from 07:04:22UTC to 07:45:27UTC, due to a power outage, the server was not
available from the Internet (the server worked well during this time interval).

(11/06/2006) - Importance:

During the time interval from 01:44:16UTC to 05:03:03UTC, due to a UPS failure at one of our Intranet nodes,
caused by a mains blackout, the server was not available from the Internet (the server worked well during
this time interval).

(24/03/2006) - Importance:

During the time interval from 05:00:00UTC to 14:00:00UTC our NOC will be down due to a major power system
maintenance task in the building where it is located. In this interval all the services will be unavailable.
Post event information: The server was unavailable from 08:15UTC to 13:05UTC.

(30/01/2006) - Importance:

The last major upgrade to our NTP servers took place. As a result, the IP address of this server changed from
80.96.120.249 - ntp3.usv.ro to 80.96.120.253 - ntp1.usv.ro as the hardware remain the same.
The main synchronization source is a Garmin 18LVC receiver with PPS output.
The operating system for this server is now FreeBSD 5.4.
During this operation the server was unavailable for about 60 minutes.

(30/12/2005) - Importance:

A major hardware upgrade to our NTP servers took place. As a result, the IP address of this server changed from
80.96.120.253 - ntp1.usv.ro to 80.96.120.249 - ntp3.usv.ro as the hardware remain the same.

(23/11/2005) - Importance:

An upgrade to the GPS receiver software (GPS 18USB) from version 2.60 to 2.80 was really helpful.
The Garmin.com developers corrected a bug described as "Fixed NMEA timestamp when unit reports a lost fix".
The effect is really nice, as in the time periods when the fix is lost, the unit reports a "good" time. The logs
show a great server time sync improvement with a very low drift (+/- 1 ... 3 ms) from UTC (with 2.60 firmware
the drift was +/- 5 ... 40 ms and much more in the interval with no position fix).
During this operation the server was unavailable for about 45 minutes.

(21/11/2005) - Importance:

We made an upgrade to our local network Ethernet connection. The connection between the laboratory
where the time servers are located and the usv.ro backbone is done via Gigabit Ethernet links. Two
of the three Ethernet cards (in ntp2.usv.ro and ntp3.usv.ro) are now Gigabit ones. In the near future we plan
to upgrade the network card in ntp1.usv.ro (due to uptime considerations).
During this operation the server was unavailable for a very short period (under 30 minutes).

(20/11/2005) - Importance:

Due to extreme weather condition (heavy snow and low temperature) we had to improve the external GPS
housing by mounting a small lamp inside the box (only 5W/12V). After this small modification, we think the GPS
receiver will never be disturbed by show.
During this operation the server availability was not affected.

(30/08/2005 - 01/09/2005) - Importance:

The time server will not be accessible from the Internet due to the reorganization of our network access point.

(15, 16 and 17 August 2005) - Importance:

Due to several power outages the server was not reachable from the Internet (even the DNS server was down,
and the name could not be resolved) . There were multiple several hours mains interruptions, beyond our control.
Under normal working conditions (respecting the EN 50160:2000 standard), these problems should not appear.

NOTICE:  This server will never be a pool.ntp.org member !

WARNING : In order to use this server you MUST send an Email to ecoca-at-usv.ro BEFORE configuring your client software !
Failing to do this will have as a result your appearance in a "RESTRICT/DENY" line in ntpd.conf or, more probably, in ipfw.rules.

Server administrator: Eugen COCA - www.ecoca.ro