- ripe-atlas - mailman.ripe.net

Strange and undocumented (?) result in traceroute JSON output
by Stephane Bortzmeyer 17 Jul '13

17 Jul '13

Measurement #1013405 contains: "result": [ { "x": "*" }, { "late": 20, "ttl": 61, "from": "88.79.15.141", "size": 96 }, { "x": "*" }, { "x": "*" } While "late" is documented <https://atlas.ripe.net/doc/data_struct#v4400_traceroute>, the lack of "rtt" field is not and crashes tools like json2traceroute.py.

3 2

[UDM] What value to give to 'requested' when 'type' = 'msm'?
by Stephane Bortzmeyer 17 Jul '13

17 Jul '13

The documentation <https://atlas.ripe.net/doc/measurement-creation-api/> says: "probes": [ { "requested": 1, "type": "msm", "value": 1000002 } ] but, from my tests, if you do so, you get only one probe, whatever measurement #1000002 actually used. To get all the probes, I had to indicate a very large number in 'requested' (unless you remember or count the actual number).

2 2

IPv6 and Atlas probe v3
by Mike. 16 Jul '13

16 Jul '13

My probe works fine on IPv4, obtaining IP address, router, and DNS servers via DCHP. The IPv6 capability has me curious with two questions... I statically configured the IPv6 address via the web interface. I gave the probe an IPv6 address from my prefix, a gateway address, and the IP address of a DNS recursive server provided by my tunnel vendor. The IPv6 address shows up OK on the web interface ("Internet Address") The AS Number also shows up fine. The Gateway is "Undetermined/Unknown" The IP addr of the DNS server is listed as shown below. In the text table below the configuration info, I see: ("V" is a check, "X" is an X) Static Enabled Probe Knows Probe Uses Last Static Address Given IPv6 V V V 2001:470:xxx:xxx::xxx DNS V V X 2001:470:20::2 Question #1: Why is there an "X" for DNS in the "Probe Uses" column? Is there something wrong with that DNS server? When I query DNS from another box on the same subnet as the Atlas probe I get valid results: ; <<>> DiG 9.8.3-P4 <<>> @2001:470:20::2 atlas.ripe.net any ; (1 server found) ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 19739 ;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 0 ;; QUESTION SECTION: ;atlas.ripe.net. IN ANY ;; ANSWER SECTION: atlas.ripe.net. 5922 IN AAAA 2a01:4f8:121:30a3::78:16 atlas.ripe.net. 5911 IN A 178.63.78.16 ;; Query time: 21 msec ;; SERVER: 2001:470:20::2#53(2001:470:20::2) ;; WHEN: Mon Jul 8 12:24:01 2013 ;; MSG SIZE rcvd: 76 I also noticed this sentence at the bottom of the https://atlas.ripe.net/doc/static-config page: "...As of now, we have no way of preventing the probe from accepting RAs. So if there is an IPv6 (rogue) RA on you network, it can/will override part of the static configuration. We're unsure if this will change in the future. ..." I do have RA support on the subnet, though I would not classify it as "rogue". :) The RA provides the IPv6 prefix, prefix length, and the IPv6 address of the recursive DNS server. This RA support has been used by another box on the subnet, and it seems to work fine. Question #2: Does the Atlas v3 probe use RA if they are present? Thanks.

2 2

Just to say thank you
by Stephane Bortzmeyer 16 Jul '13

16 Jul '13

With Atlas probes, we were able (at the cost of a few credits) to solve a difficult network connectivity problem. I just wanted to say THANKS.

1 0

Re: [atlas] Timeout on Ping Measurements
by Philip Homburg 15 Jul '13

15 Jul '13

On 2013/07/14 19:35 , Rusty Dekema wrote: > > What is the default timeout on ICMPv4 user-defined measurements made > on RIPE Atlas? I am guessing something like 1.2 s based on > observations, but I'd like to know definitively. It should be 1 second. > Also, is it possible to increase the timeout? One of the targets I > have been measuring often takes a very long time to respond (due to > congestion or link-layer retries) and I would like to have a high (5 > second, perhaps) timeout on these pings. I haven't noticed an option > to set this when setting up the measurement, but it's very possible > I've overlooked something. At the moment there is no option to change the ping timeout. Traceroute does have that option. Philip

1 0

Timeout on Ping Measurements
by Rusty Dekema 14 Jul '13

14 Jul '13

Greetings, What is the default timeout on ICMPv4 user-defined measurements made on RIPE Atlas? I am guessing something like 1.2 s based on observations, but I'd like to know definitively. Also, is it possible to increase the timeout? One of the targets I have been measuring often takes a very long time to respond (due to congestion or link-layer retries) and I would like to have a high (5 second, perhaps) timeout on these pings. I haven't noticed an option to set this when setting up the measurement, but it's very possible I've overlooked something. Cheers, Rusty Dekema

1 0

[iesg-secretary@ietf.org: [lmap] WG Action: Formed Large-Scale Measurement of Broadband Performance (lmap)]
by Stephane Bortzmeyer 11 Jul '13

11 Jul '13

It seems quite close from what Atlas is doing. Atlas documentation in a RFC, one day? :-)

2 1

300 probes just disappeared?
by Stephane Bortzmeyer 11 Jul '13

11 Jul '13

Yesterday, on the Web site, there were > 3400 probes, now: Probes connected to RIPE Atlas: 3177

3 2

atlas.ripe.net website reliability
by Mike. 11 Jul '13

11 Jul '13

About 20% of the time I try to access the website, the browser times out. Granted, I tend to access the website when it is the early morning hours in Europe, perhaos maintenance is being performed. Are there any plans in place to make the website more available to global viewers?

5 6

A Tutorial on Large-Scale Measurement Platforms
by Vaibhav Bajpai 03 Jul '13

03 Jul '13

Hello, We recently gave a tutorial on Large-Scale Measurement Platforms at AIMS 2013 [1]. The tutorial is available at [2]. [1] http://www.aims-conference.org/2013/labs.html [2] http://cnds.eecs.jacobs-university.de/tutorials Abstract: --------- This tutorial provides an introduction into large-scale measurement platforms. It begins by surveying active measurement studies that paved way for the deployment of such platforms. It then describes the internals of the RIPE Atlas and SamKnows platforms, their architecture and surveys the work published from insights discovered from the collected data. The first part concludes by introducing the current Large Scale Measurement of Access Network Performance (LMAP) standardisation efforts in the IETF. The second part of the tutorial provides practical training on how to virtualise an OpenWrt-based Customer Premises Equipment (CPE) and bootstrap it to register as a Measurement Agent (MA) with a controller. The MA is then used to install a measurement test, schedule the reporting of the measurement test results and use a REST API to send and retrieve the measurement results back from the collector. The tutorial also provides training on how to use the RIPE Atlas REST API to create User Defined Measurements (UDMs), and retrieve the UDM measurement results. It concludes with a description of the UDM result data structures and a brief of how programming tools can be used to analyse such data. Best Regards, Vaibhav Bajpai and Nikolay Melnikov About authors: Vaibhav Bajpai is a PhD student at Jacobs University Bremen, Germany. His research focuses on understanding the impact of network infrastructure changes using Large-Scale Measurement Platforms. http://vaibhavbajpai.com Nikolay Melnikov is a PhD student at Jacobs University Bremen, Germany. His research focuses on identification and differentiation of Internet users based on their browsing patterns and behaviors. http://nmelnikov.com

1 0