...
The output appears as:Code Block cottrell@pinger $ cp /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2006-09-28.txt.gz /tmp/ cottrell@pinger $ zcat /tmp/minimum_rtt-100-by-node-2006-09-28.txt.gz cottrell@pinger $ head /tmp/minimum_rtt-100-by-node-2006-09-28.txt 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 pinger.slac.stanford.edu manila.global.net.pg 547.877 548.021 547.934 548.611 548.680 548.069 549.023 548.124 548.687 547.573 548.639 549.181 547.998 548.302 547.858 547.464 548.470 548.251 547.849 547.708 548.463 547.473 547.651 547.732 pinger.slac.stanford.edu manila.global.net.pg pinger.slac.stanford.edu lobelia.physics.wisc.edu 61.089 60.958 61.037 61.332 60.958 61.262 61.073 61.093 61.040 61.086 60.974 60.979 60.956 61.070 60.942 61.035 61.994 61.022 60.927 60.957 61.091 61.009 61.024 60.974 pinger.slac.stanford.edu lobelia.physics.wisc.edu
By default the above file is created once thus the directory appears as:
Code Block 57cottrell@pinger:~>ls -l /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05* -rw-rw-r-- 1 pinger iepm 492144 May 2 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-01.txt.gz -rw-rw-r-- 1 pinger iepm 545968 May 3 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-02.txt.gz -rw-rw-r-- 1 pinger iepm 561661 May 4 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-03.txt.gz -rw-rw-r-- 1 pinger iepm 566550 May 5 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-04.txt.gz -rw-rw-r-- 1 pinger iepm 537127 May 6 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-05.txt.gz -rw-rw-r-- 1 pinger iepm 538830 May 7 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-06.txt.gz -rw-rw-r-- 1 pinger iepm 488360 May 8 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-07.txt.gz -rw-rw-r-- 1 pinger iepm 499020 May 9 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-08.txt.gz -rw-rw-r-- 1 pinger iepm 563840 May 10 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-09.txt.gz -rw-rw-r-- 1 pinger iepm 583454 May 11 02:17 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-10.txt.gz -rw-rw-r-- 1 cottrell iepm 577949 May 12 22:08 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-11.txt.gz -rw-rw-r-- 1 cottrell iepm 102 May 12 17:25 /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-100-by-node-2011-05-12.txt.gz
Example output format. Following the 1st line in the file there is 1 line like the following per day/per host pair. Between the initial and final src_name and tgt_name tokens there are 24 tokens one for each hour of the day, missing data is identified by a dot followed by a space (. ), e.g.:
Code Block icfamon.dl.ac.uk lns62.lns.cornell.edu 108.871 . . . . . . . . 108.892 . . . . . . . . . . . . . 109.620 icfamon.dl.ac.uk lns62.lns.cornell.edu
The first line in the file contains a label for each of the time slots (e.g. hours):Code Block 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Example output filename for the minimum_rtt metric:The remaining analyze scripts (wrap-analyze-daily.pl, wrap-analyze-monthly.pl, wrap-analyze-allmonths.pl, and wrap-analyze-allyears.pl) take as input the data from wrap-analyze-hourly.pl, wrap-analyze-daily.pl, and wrap-analyze-allmonths.pl and create files of the form:
Code Block /nfs/slac/g/net/pinger/pingerreports/hep/<metric>-<size>-by-<site|node>(-<YYYY>?)(-<mm>?)(-<dd>?).txt.gz /nfs/slac/g/net/pinger/pingerreports/hep/<metric>-<size>-by-<site|node>-<60|120|365>days.txt.gz /nfs/slac/g/net/pinger/pingerreports/hep/<metric>-<size>-by-<site|node>-<allmonths|allyears>.txt.gz
There are ~ 16 metrics:
<option value="MOS">Mean Opinion Score</option> <option value="alpha">Directivity</option> <option value="average_rtt" selected>Average Round Trip Time</option> <option value="conditional_loss_probability">Conditional Loss Probability</option> <option value="duplicate_packets">Duplicate Packets</option> <option value="ipdv">Inter-Packet Delay Variation</option> <option value="iqr">Inter-Quartile Range</option> <option value="maximum_rtt">Maximum Round Trip Time</option> <option value="minimum_packet_loss">Minimum Packet Loss</option> <option value="minimum_rtt">Minimum Round Trip Time</option> <option value="out_of_order_packets">Out of Order Packets</option> <option value="packet_loss">Packet Loss</option> <option value="throughput">TCP Throughput (kbits/s)</option> <option value="unpredictability">Ping Unpredictability</option> <option value="unreachability">Ping Unreachability</option> <option value="zero_packet_loss_frequency">Zero Packet Loss Frequency</option> Information on these can be found at http://www.slac.stanford.edu/comp/net/wan-mon/tutorial.html.
- More information can be found on running getdata.pl and the analyze scripts to recover missing data.
- PRM
- /afs/slac/package/pinger/prm/prm.pl
- The script prm.pl, is used to create csv files for the analyzed data which can then be used by PingER motion charts or PingER Executive Plots.
The script requires a configuration file, which contains entries for all the reports prm should create. These entries are of the form:
Code Block <metric name> <monitoring site> <country||continent> <tick> (<filter>?)
- It gets its data from pingtable.pl either by calling it locally or by accessing it via the web.
- The reports created by prm are available online.
- The reports are used by the metric motion plots, the metric maps and the executive plots.
An example of the output is seen below:
allyearly,?,1998,1999,2000,2001,2002,2003,2004,2005,2006,2007,2008,2009,2010
EDU.SLAC.STANFORD.N3-to-Afghanistan,.,.,.,.,.,.,.,.,.,.,12.504,4.481,9.257,5.284
EDU.SLAC.STANFORD.N3-to-Albania,.,.,.,.,.,.,.,.,.,.,.,4.949,7.133,4.590
later it is atomically copied to files in the permanent directory
A period(.) means there was no data.
There is a directory with lots of these analyses, see https://www-iepm.slac.stanford.edu/pinger/prmout/
Order of running jobs in trscrontab
...
Data
Data flow
The raw data is measured by pinger2.pl and gathered from the MAs using getdata_all.pl
The hourly data is created from the raw data by /afs/slac/package/pinger/
...
analysis/
...
...
and saved in
/nfs/slac/
...
g/net/pinger/pingerreports/hep/<metric>/<metric>-<sizte>-by-<host|site>-<yyyy>-<mm>.txt.gz, where <yyyy> is the 4 digit year (e.g. 2021), and <mm is the 2 digit (including leading zero)> month (e.g. 02).
| Code Block |
|---|
/nfs/slac/g/net/pinger/pinger_mon_data/ping-<YYYY>-<MM>.txt contains raw data from 2005-2010, just for SLAC.
/nfs/slac/g/net/pinger/pinger2/data/ping-<YYYY>-<MM>.txt contains raw data from 2009-present, just for SLAC
/nfs/slac/g/net/pinger/pingerdata/<data> contains data from 1997-2003 contains raw data
/nfs/slac/g/net/pinger/pingerdata/hep/data/<host>/ contains raw data 1997-200 |
Data volumes Sep 2014
The total PingER data volume is about 550Gbytes.
We estimate that there are about 60 GBytes of uncompressed hourly data for 100 Byte pings by node, as of September 2014. One would estimate to about quadruple that if one added 1000 byte pings and by site. See Volume of PingER data Sep 2014.
Archiving
See Archiving PingER data by tar for retrieval by anonymous ftp
Missing data
Anomalies
We have spotted anomalies between the values reported by:
- table.pl and pingtable.pl
- and pingtable.pl at SLAC and pingtable.pl at SEECS
They are discussed and explained in the Anomalies report.
Problems
I get the error message below:
| Code Block |
|---|
Your "cron" job /nodes.cf #Takes about 30 secs # Create the offsite.nodes list of nodes to ping from Guthrie NODEDETAILS database, add by Cottrell 1/3/2107 takes ~25 minutes # The following files are created: # /afs/slac.stanford.edu/www/comp/net/mon/offsite.nodes #txt file of all node, format: name address #comment # /afs/slac.stanford.edu/g/www/www-iepm/pinger/beacons.txt #txt file of Beacons # /afs/slac.stanford.edu/g/www/www-iepm/pinger/pinger.xml #xml file for non SLAC MA # /afs/slac.stanford.edu/package/pinger/pinger2/share/pinger/pinger.xml #xml file for SLAC MA lnxcron 00 23 * * * /afs/slac/package/pinger/write_offsitenodes.pl > /afs/slac/g/www/www-iepm/pinger/host-warnings.txt 2>&1 #takes ~15 minutes on Pinger #--------------Normal complete set for getdata.pl---------------------- pinger;200 32 1 * * * /afs/slac/package/pinger/getdataanalysis/wrap-analyze-daily.pl --basedir > /afsnfs/slac/g/www/www-iepmnet/pinger --usemetric --dataset hep --by by-site --size 1000 produced the following output: Thu Dec 3 05:00:02 2015 Warning /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-1000-by-site-2015-12-01.txt.gz does not exist |
Since the trscronjob analyze-all.pl --date 1days only reads and analyzes the most recent day's raw data if the job fails to run, then next days there will be missing data and you wilr get the above message. To recover the missing daily data run analyze-all.pl --date 2015-12-01 for the missing date (in this example 2015-12-01).
Backup
On 3/9/2012, we requested unix-admin@slac.stanford.edu to backup /nfs/slac/g/net/pinger/ on a regular basis. This was added to the nightly backup by Andrew May.
Order of running jobs in trscrontab
| Code Block |
|---|
##############Basic PingER files scripts follow ###################### #Update nodes.cf (Perl dump of Guthrie config) from the NODEDETAIL Guthrie database, add by Cottrell 1/31/07 Though node.pl is called from runall.pl, runall.pl is not called until 5am, so it is updated here first to be sure.It takes about 3 seconds to run pinger 30 0,6,12,18 * * * /afs/slac/package/pinger/guthrie/node.pl -o > /slaconly/getdata.err #Takes ~20 minutes 7/31/2019 #Minimal subset is /afs/slac/package/pinger/getdata_all.pl -h pinger.slac.stanford.edu #Takes < 1 minute for pinger.slac.stanford.edu #-----------------analyze-hourly.pl runs first for both 100B & 1000B pings, by-host, by site-------------------------- lnxcron;120 55 2 * * * /afs/slac/package/pinger/analysis/analyze-all.pl --date 1days #Takes 25:14 10/20/2011 (55 mins 9/21/2011, 70 minutes 5/11/2018) #-----------------analyze-daily------------------------------------------- pinger;300 30 4 * * * /afs/slac/package/pinger/analysis/wrap-analyze-daily.pl --basedir /nfs/slac/g/net/pinger --use/metric --dataset hep # Takes ~ 5 hours for whole month on pinger 11/10/2012 #------------------analyze-monthly.pl for defaults by-node, size=100B takes about 60 minutes-------------- lnxcron 12 10 * * * /usr/local/bin/bsub -W 200 -o /dev/null -q xxl /afs/slac/package/pinger/analysis/wrap-analyze-monthly.pl --basedir /nfs/slac/g/net/pinger --usemetric --dataset hep > /dev/null #Takes about 55mins 9/19/2011, 61 mins 10/9/2019 on pinger or 7 mins on rhel6 11/13/2011.nodes.cf #Takes about 30 secs # Create the offsite.nodes list of nodes to ping from Guthrie NODEDETAILS database, add by Cottrell 1/3/2107 takes ~25 minutes # The following files are created: # /afs/slac.stanford.edu/www/comp/net/mon/offsite.nodes #txt file of all node, format: name address #comment # /afs/slac.stanford.edu/g/www/www-iepm/pinger/beacons.txt #txt file of Beacons # /afs/slac.stanford.edu/g/www/www-iepm/pinger/pinger.xml #xml file for non SLAC MA # /afs/slac.stanford.edu/package/pinger/pinger2/share/pinger/pinger.xml #xml file for SLAC MA lnxcron 00 23 * * * /afs/slac/package/pinger/write_offsitenodes.pl > /afs/slac/g/www/www-iepm/pinger/host-warnings.txt 2>&1 #takes ~15 minutes on Pinger #------------------sites-per-country-----------------Normal complete set for getdata.pl---------------------- pinger;200 0032 091 * * * /afs/slac/package/pinger/sites-per-countrygetdata.pl #Takes anout 10 minutes, run after analyze monthly |
There is a script /u/sf/cottrell/bin/manual-upd8-sites-per-country.pl that will run in sequence the minimal set of scripts needed to update sites-per-country.html.
Checks:
- View pinger.xml at SLAC
- View log of measurements from SLAC, or view form
- View gathered data for SLAC at /nfs/slac/g/net/pinger/pingerdata/hep/data/pinger.slac.stanford.edu/ping-<YYYY>-<MM>-<DD>.txt.gz
- View analyzed hourly data
- View analyzed daily data
- View analyzed monthly data
- View sites per country
Data
| Code Block |
|---|
/nfs/slac/g/net/pinger/pinger_mon_data/ping-<YYYY>-<MM>.txt contains raw data from 2005-2010, just for SLAC.
/nfs/slac/g/net/pinger/pinger2/data/ping-<YYYY>-<MM>.txt contains raw data from 2009-present, just for SLAC
/nfs/slac/g/net/pinger/pingerdata/<data> contains data from 1997-2003 contains raw data
/nfs/slac/g/net/pinger/pingerdata/hep/data/<host>/ contains raw data 1997-200 |
Data volumes Sep 2014
The total PingER data volume is about 550Gbytes.
We estimate that there are about 60 GBytes of uncompressed hourly data for 100 Byte pings by node, as of September 2014. One would estimate to about quadruple that if one added 1000 byte pings and by site. See Volume of PingER data Sep 2014.
Archiving
See Archiving PingER data by tar for retrieval by anonymous ftp
Missing data
Anomalies
We have spotted anomalies between the values reported by:
- table.pl and pingtable.pl
- and pingtable.pl at SLAC and pingtable.pl at SEECS
They are discussed and explained in the Anomalies report.
Problems
I get the error message below:
| Code Block |
|---|
Your "cron" job
/afs/slac/package/pinger/analysis/wrap-analyze-daily.pl
--basedir /nfs/slac/g/net/pinger --usemetric --dataset hep --by by-site --size 1000
produced the following output:
Thu Dec 3 05:00:02 2015 Warning /nfs/slac/g/net/pinger/pingerreports/hep/minimum_rtt/minimum_rtt-1000-by-site-2015-12-01.txt.gz does not exist |
Since the trscronjob analyze-all.pl --date 1days only reads and analyzes the most recent day's raw data if the job fails to run, then next days there will be missing data and you wilr get the above message. To recover the missing daily data run analyze-all.pl --date 2015-12-01 for the missing date (in this example 2015-12-01).
Backup
...
> /afs/slac/g/www/www-iepm/pinger/slaconly/getdata.err #Takes ~20 minutes 7/31/2019
#Minimal subset is /afs/slac/package/pinger/getdata_all.pl -h pinger.slac.stanford.edu #Takes < 1 minute for pinger.slac.stanford.edu
#-----------------analyze-hourly.pl runs first for both 100B & 1000B pings, by-host, by site--------------------------
lnxcron;120 55 2 * * * /afs/slac/package/pinger/analysis/analyze-all.pl --date 1days #Takes 25:14 10/20/2011 (55 mins 9/21/2011, 70 minutes 5/11/2018)
#-----------------analyze-daily-------------------------------------------
pinger;300 30 4 * * * /afs/slac/package/pinger/analysis/wrap-analyze-daily.pl --basedir /nfs/slac/g/net/pinger --use/metric --dataset hep # Takes ~ 5 hours for whole month on pinger 11/10/2012
#------------------analyze-monthly.pl for defaults by-node, size=100B takes about 60 minutes--------------
lnxcron 12 10 * * * /usr/local/bin/bsub -W 200 -o /dev/null -q xxl /afs/slac/package/pinger/analysis/wrap-analyze-monthly.pl --basedir /nfs/slac/g/net/pinger --usemetric --dataset hep > /dev/null #Takes about 55mins 9/19/2011, 61 mins 10/9/2019 on pinger or 7 mins on rhel6 11/13/2011.
#------------------sites-per-country---------------------------------------
pinger 00 09 * * * /afs/slac/package/pinger/sites-per-country.pl #Takes anout 10 minutes, run after analyze monthly |
There is a script /u/sf/cottrell/bin/manual-upd8-sites-per-country.pl that will run in sequence the minimal set of scripts needed to update sites-per-country.html.
Checks:
- View pinger.xml at SLAC
- View log of measurements from SLAC, or view form
- View gathered data for SLAC at /nfs/slac/g/net/pinger/pingerdata/hep/data/pinger.slac.stanford.edu/ping-<YYYY>-<MM>-<DD>.txt.gz
- View analyzed hourly data
- View analyzed daily data
- View analyzed monthly data
- View sites per country
Accessing PinGER data
| Include Page | ||||
|---|---|---|---|---|
|
...