Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

On EOL systems, the default value of the DATADIR environment variable should be "merge", in which case statsproc will process all files on /scr/isfs/projects/CABL/merge.  If you want to process a different set of files, for example you can pass the list of files instead of the start and end time.  For example, the 50m files:

cd /scr/isfs/projects/CABL/raw_data

...

rm $ISFF/projects/CABL/ISFF/logs/R_derived_last.txt

Time keeping

The DSMs each have a GPS with a pulse-per-second signal. Using NTP reference clock software, each DSM is then a stratum 1 time server.  NTP on the DSM uses the GPS reference clock to adjust the CPU system clock, and generally reports that the GPS reference clock has less than a 50 micro-second offset from the system clock.

To query the system clock on a tower DSM from flux, use the ntpq -p command, for example 50m:

ntpq -p 50m
remote refid st t when poll reach delay offset jitter
==============================================================================
LOCAL(0) .LOCL. 10 l 20d 64 0 0.000 0.000 0.000
oGPS_NMEA(0) .GPS. 0 l 5 16 377 0.000 -0.001 0.031

The above shows the GPS reference clock is offset from system CPU clock by -0.001 milliseconds. The "reach" value for GPS_NMEA should be 377 (octal value of all 1's). The reach for the LOCAL clock is always 0.

flux is configured to use all 6 DSMs as network time servers, using chrony, a NTP client. To display the current chrony status, use chronyc sources:

chronyc sources
210 Number of sources = 6
MS Name/IP address Stratum Poll Reach LastRx Last sample
===============================================================================
^+ 50m                                 1  10     377     558    -298ns[+5000ns] +/- 742us
^+ 100m                               1  10     377     179     +52us[    +52us] +/- 662us
^+ 150m                               1  10     377     219 +2000ns[+2000ns] +/- 739us
^* 200m                                1  10     377     338 +9656ns[    +15us] +/- 671us
^+ 250m                               1  10     377     439 +5677ns[    +11us] +/- 732us
^+ 300m                               1  10     377       95     +11us[    +11us] +/- 794us

The above shows that the clock on flux agrees to within a maximum 52 microseconds with each DSM, as indicated by the "Last sample" value in brackets, which is the last calculated offset of the reference clock (in this case the DSM) from the system time on flux. A positive offset means the reference clock is ahead of the system clock.

The second character, under "S" should be '*' (indicating chrony on flux is sync'd to this server) or '+' (good server).  You may also see '-' (recently on 100m for some reason) indicating chrony does not have a high opinion of its time information, relative to the others.

The "reach" values should again be 377. If not, it means the DSM is not on the network, or its NTP server is not responding or sync'd to its GPS.

One can also ssh into bao or ehs and check their clocks with "ntpq -p".