mxmlnkn's questions

I have written a program example.jar which uses a spark context. How can I run this on a cluster which uses Slurm? This is related to https://stackoverflow.com/questions/29308202/running-spark-on-top-of-slurm but the answers are not very detailed and not on serverfault.

For educational purposes I wanted to trace a chain of NTP servers, e.g. 0.de.pool.ntp.org back to a stratum 1 NTP server. How can I do this?

I found ntptrace, but it doesn't work:

/home/xyzdragon# ntptrace
localhost: stratum 2, offset -0.009285, synch distance 0.010221
192.53.103.104: timed out, nothing received
***Request timed out

I tried to debug /usr/bin/ntptrace by reproducing the steps of that Perl-script manually:

home/xyzdragon# ntpq -n
ntpq> pe
     remote           refid      st t when poll reach   delay   offset  jitter
==============================================================================
+78.47.249.19    56.1.129.236     3 -  129  128  376   27.339   14.405  12.857
ntpq> host 78.47.249.19
current host set to 78.47.249.19
ntpq> pe
78.47.249.19: timed out, nothing received
***Request timed out

Actually ntptrace uses a combination of rv and pstat instead of the pe ntpq command.

After several tries getting it to work with ntpq, I just used a workaround:

1. Read peer and refid from `ntpq -p
2. Insert server 56.1.129.236 (skipping 78.47.249.19) into /etc/ntp.conf
3. Apply settings with sudo service ntp restart
4. Go to 1.

This works to trace an NTP chain, but it is very tedious. My questions therefore are:

• How can I use maybe ntpq only to trace the NTP server chain?
• Why isn't ntptrace working in the first place?