List of activities for pool of time servers at

2024-04-21: Statictical data for cluster.

On world wide scale cluster serves 0.052% of NTP traffic from NTP pool. In other words cluster serves every 1/1923th of NTP packet on world. This is not a huge percentage but still presents more than 1640 packets per second. Percentages are better for IPv6 since there is less IPv6 NTP serves, but supprisingly cluster serves packets to Brazil and China IP address space as well. Cluster resides in Slovenia so dominance from Slovenina IP address space is natural. Based on number of NTP severs in different regions, this cluster of four servers is doing pretty well. Servers in cluster serve on average 2.3 times more packets per second than average NTP server. This is probbaby because all servers in cluster are stratum one servers, and they have better chance for selection. NTP cluster statistical data on 2024-04-24 15:00 UTC

2024-02-09: Network provider service disruption.

Today, 2024-02-09, we had an internet service disruption from our network provider T-2. The disruption lasted from 11:32 CET to 12:58 CET (from 10:32 UTC to 11:58 UTC). During this time, all network NTP services were unavailable. According to preliminary news, something went wrong at the T-2 network operator electrical substation. At the time of writing this article (2024-02-09 at 16:35 CET), all servers are up and running, but not all are yet promoted to full operation from the NTP pool due to a low score. The expected normal operation of service, with minimum score 10, will be by 2024-02-09, 17:15 CET.

No electricity 2024-02-09

2023-10-07: Server load packets/s

Many would like to know what is typical load of one NTP server in cluster. Typical average value is 410 queries per seconds for per cluster member. That sums to around 1640 queries per seconds for whole cluster. As presented in previous post numbers slightly fluctuate across day and week day. Graph on rihgt represent one week of average requests per seconds in roughly 20 minutes interval. It shows time distribution of around 248 million queries during one week time interval.

NTP v4 queries per second in 20 minutes interval for one cluster member

2023-08-27: Bandwith Increase for cluster

After upgrade to chrony 4.4 we also increased bandwidth on 2023-08-13 for each time server. It took around 10 days to utilize bandwidth to current usage of 224 Kbits/s out of reserved 10Mbit for each time server. As result number of NTP packets increase to 353 packets/s for each time server and on average 1413 requests/s for cluster. We monitor all traffic parameters, details are in table: IPv4 and IPv6 interface statistic

Increased chrony bandwith 2023-08-14

2023-08-10: Upgrade to chrony 4.4.

On 2023-08-09 chrony time server received another upgrade to version 4.4. We upgraded complete cluster with chrony 4.4 from source. Official chrony web page moved to new url which is now:'

Upgrade to chrony 4.4

2023-01-12: Upgrade to gpsd 3.25.

On 2023-01-25 gpsd service daemon received another upgrade to version 3.25. We upgraded complete cluster with gpsd 3.25 package compiled from source.

Upgrade to gpsd 3.25

2022-11-01: NTS secured time distribution.

After successfull testing period we upgraded complete cluster with Network Time Security (NTS). The NTS mechanism uses Transport Layer Security (TLS) to establish the keys needed for authentication of NTP packets. Complete protocol definition is specified in: Protocol and Algorithms Specification RFC 5905, Network Time Protocol Version 4 Extension Fields RFC 7822, Message Authentication Code for the Network Time Protocol RFC 8573, Network Time Protocol Version 4 Port Randomization RFC 9109. Certificates are generated with help of letsencrypt servers and updated with certbot client.

Enabled chronyd secure NTS protocol.

2022-09-01: Upgrade of gpsd and chrony packages.

Complete cluster has been upgraded with gpsd version 3.24 and chronyd server version 4.3 packages. This update consists of custom configured and optimization compiled source code for greater accuracy and stability. Check gpsd web site and chrony web site .

New gpsd v3.24 and chronyd v 4.3 releases.

2022-05-14: Timing modules upgrade on location 2.

After general shortage of electronic we finally manage to upgrade time servers on location 2. We use same timing modules accross all four time servers. This simplifies integration of HW equipment. Now we run two servers on location 2 in test mode to ensure future smooth operation. From 2022-06-01 we run all four time servers in full production mode.

Timing modules upgrade on location 2.

2022-01-06: Dedicated link bandwith.

We dedicate on each location internet connection with symetrical bandwith of 10Mbit/s per pair of time servers. Currentlly is consumed about in: 108 Mbit/s and out: 126Mbit/s per time server (in: 220 Mbit/s and out: 250 Mbit/s per location). This ensures us smooth operation of all time servers. With all available resources we have room for future growth.

Bandwidth consumed on one time server.

2021-08-31: Planned upgrade on location 2.

We have plan to upgrade time servers on location 2 to stratim one. Expected time frame is by end of December 2021. Location 2 servers will be upgraded to same HW as on location 1, to stratum 1 time servers. This will assure that we will have two fail over locations each with two stratum 1 servers linked in active-active fail over configuration.

Planned server upgrade on location 2

2021-08-01: New web site.

We operate public time servers since 2018, but now we decide to more actively participate with time services to public community. After preparation period web site is alive. Main objective of web site is to show how to configure NTP service on different operating systems and to show current status of pool monitored time servers. At start we have two locations in Slovenia / Europe. On location 1 we have two stratum 1 time servers in active/active fail over configuration. On location 2 we have two stratum 2 time servers in active/active fail over configuration on virtual machines.

Initial web site