Choose the fastest server automatically
netselect [\|-v\||\|-vv\||\|-vvv\||\|-vvvv\|] [\|-m HOPS] [\|-s SERVERS\|] [\|-t PACKETS\| [\|-I] [\|-D] host...
netselect determines several facts about all of the hosts given on the command line, much faster than you would if you manually tried to use ping and traceroute.
For each host, netselect figures out the approximate ping time (though not as accurately as ping does), the number of network "hops" to reach the target, and the percentage of ping requests that got through successfully. Then netselect calculates the "score" of each operational host based on these values. A lower score is better, in the end it prints one line showing the server with the best score.
If the DNS name has multiple IP addresses associated with IT netselect will test each of the IP addresses independently. In this case, netselect will report the IP address of the fastest server (unless the -D option is used).
If the number of found hosts is not the same as the number requested (using -s), netselect will emit a warning message. If no hosts are found at all, the program will also end with an error return value.
# netselect -vv ftp.fceia.unr.edu.ar ftp.kulnet.kuleuven.ac.be \ ftp.cdrom.com ftp.debian.org ftp.de.debian.org
This is the output:
ftp.fceia.unr.edu.ar 2792 ms 23 hops 100% ok ( 1/ 1) [ 9213] ftp.kulnet.kuleuven.ac.be 9999 ms 30 hops 0% ok ftp.cdrom.com 94 ms 8 hops 100% ok (10/10) [ 169] ftp.debian.org 46 ms 15 hops 100% ok (10/10) [ 115] ftp.de.debian.org 9999 ms 30 hops 0% ok 115 ftp.debian.org
The value in brackets is the "score" of each operational host based on these values. A lower score is better. The last line shows the server with the best score. If we had not used '-vv' on the command line, only this last line would have been printed.
Note that for ftp.kulnet.kuleuven.ac.be and ftp.de.debian.org in this case, nothing got through at all. That indicates that either the host doesn't exist, or it is down.
# netselect -vv http.us.debian.org
This is the output:
204.152.191.39 300 ms 17 hops 90% ok ( 9/10) [ 899] 35.9.37.225 9999 ms 30 hops 0% ok 64.50.233.100 9999 ms 30 hops 0% ok 128.30.2.36 183 ms 15 hops 90% ok ( 9/10) [ 510] 149.20.20.135 226 ms 18 hops 90% ok ( 9/10) [ 702] 510 128.30.2.36
In this case, since the single name has multiple DNS addresses all of the servers will be tested independently and the fastest server will be provided.
If you notice the above output you will see that the time for some hosts is 9999 ms. This is typically an indication of remote hosts blocking the UDP probes netselect relies on. To prevent this issue you can use ICMP tests. For example, repeating the above test using ICMP, that is:
# netselect -I -vv http.us.debian.org
yields the following output:
204.152.191.39 291 ms 17 hops 100% ok (10/10) [ 785] 35.9.37.225 180 ms 19 hops 66% ok ( 2/ 3) [ 783] 64.50.233.100 140 ms 12 hops 100% ok (10/10) [ 308] 128.30.2.36 182 ms 15 hops 100% ok (10/10) [ 455] 149.20.20.135 227 ms 18 hops 100% ok (10/10) [ 635] 308 64.50.233.100
In this last example the selected server will change since, actually, the fastest server is one that also actively blocks UDP probes.
You can also use the tag feature based on postpending the server name with itself so the result includes both the fastest IP address and the original server name. For example, running a query with tags such as this one:
# netselect -I -vv http.us.debian.org:http.us.debian.org
will yield output similar to:
35.9.37.225:http.us.debian.org 9999 ms 30 hops 0% ok 64.50.233.100:http.us.debian.org 9999 ms 30 hops 0% ok 64.50.236.52:http.us.debian.org 9999 ms 30 hops 0% ok 128.30.2.36:http.us.debian.org 189 ms 14 hops 90% ok ( 9/10) [ 504] 199.6.12.70:http.us.debian.org 286 ms 17 hops 90% ok ( 9/10) [ 858] 504 128.30.2.36:http.us.debian.org
Or you can use the -D option to force netselect to print out only DNS names. For example:
# netselect -I -vv -D ftp.us.debian.org ftp.ru.debian.org
will yield output similar to:
128.30.2.36 186 ms 14 hops 100% ok (10/10) [ 446] 199.6.12.70 294 ms 17 hops 87% ok ( 7/ 8) [ 907] 35.9.37.225 177 ms 19 hops 66% ok ( 2/ 3) [ 768] 64.50.233.100 141 ms 12 hops 100% ok (10/10) [ 310] 64.50.236.52 162 ms 13 hops 100% ok (10/10) [ 372] ftp.ru.debian.org 112 ms 18 hops 100% ok (10/10) [ 313] 310 ftp.us.debian.org
As you can see here, netselect will select a DNS name (in the example 'ftp.us.debian.org') which associated with multiple IP addresses over other servers as long as there is one server in the pool faster than others. However, using the DNS name for a server configuration might lead to actually slower throughput if, on average, the different IP addresses are not as responsive. In the above example, 'ftp.ru.debian.org' is second on the list and might be a better option (again, on average) than 'ftp.us.debian.org'.
-v
Verbose mode. Displays nameserver resolution messages to stderr. You probably want this so that you don't get bored waiting for a hundred name resolutions to finish.
-vv
Very verbose mode. Displays nameserver resolution and statistics (not just scores) to STDERR and STDOUT.
-vvv
Very very verbose mode. Everything -vv prints, plus it print every packet received as it happens. Good for debugging or trying to figure out how it works.
-vvvv
Very very very verbose mode. Everything -vvv prints, plus a trace of all packets sent.
-mHOPS
Maximum TTL (time to live). Don't accept hosts that are further than HOPS away.
-sSERVERS
Print this many "top-scoring" SERVERS at the end of the list. If SERVERS is 0, then this disables printing of high scores.
-tPACKETS
Make sure at least 50% of the hosts get tested with this many PACKETS. The more packets you use, the more accurate are the results... and the longer it takes to run. The default is 10, which is usually okay.
-I
Use ICMP instead of UDP probes. On some occasions (e.g. firewalled hosts) UDP probes like those used in traceroute will get filtered out and hosts might be perceived as dead. Using this option will make netselect use ICMP probes instead.
-D
netselect provides the IP address of the servers that are tested if the name resolves to more than one address (i.e. the remote end is using DNS round-robin). This makes it easier to spot which is the fastest server of a pool but might not be useful if you wish to configure a system daemon with the DNS name or if the remote service does not accept calls to its IP address (as some HTTP servers do). This option forces netselect to present the final server list using the original DNS name instead of the IP address if the DNS name resolves to more than one IP address.
ping(8), traceroute(8), netselect-apt(1).
Avery Pennarun <[email protected]>