Code By
Lucas Mouilleron

An implementation of heartbeat monitoring.

Definitions

• The server is in charge of monitoring services
• When a service is considered dead, the server then alerts its owner
• A service is a pulsing agent
• A pulse is a sign of aliveness
• When a service pulse the server, it tells him when he will pulse again in the worst case
• If the service has not pulsed again in time, the server considers the service dead and alerts its owner
• When a service is no longer required, it must inform the server to avoid a false dead alerts
• An Alert can be a mail, a pushover message, etc.

Implementation

• HTTP client / server architecture
• All queries protected by password set in HTTP headers under "password"
• Protocol :
  • Pulse :
    • POST /
    • {"service":"SERVICE_NAME","alertType":"ALERT_TYPE","alertTarget":"TARGET_NAME","nextIn":EXPECTED_NEXT_HEARTBEAT_IN_SECS}
  • Cancel :
    • DELETE /
    • {"service":"SERVICE_NAME"}
  • List :
    • GET /

Server

• ./server
• python3
• Dependencies : pip install -r requirements.txt
• Config :
  • config/config.json : main config (cp config/sample.json config/config.json)
  • config/certificate.pem : SSL certificate, used only if SSL activated
• Datas : data
• Deploy :
  • Install dependencies
  • Setup config
  • Generate ssl certificates (optional)
  • Hook in with upstart (optional, ./server/config/sample.upstart.conf, http://upstart.ubuntu.com/getting-started.html)
• Run : python server.py
• Docker :
  • ./docker
  • Config : place server config files in ./docker/config
  • Interactive : cd docker && tools/dockerBuild && tools/dockerRun
  • Detached : cd docker && tools/dockerBuild && tools/dockerRunDetached

Java Client

• ./clientJava
• Java 7+
• No dependencies
• Run test :

Python Client

• ./clientPython
• python3
• Dependencies : requests (pip install requests)
• Run test : python test.py

Javascript client

• ./clientJavascript
• client.js is a node module
• Dependencies : packages.json -> devDependencies (npm install)
• Run test : node test.js
• Run test in browser : browserify test.js -o bundle.js and open test.html in browser

Certificates

• Self signed :
  • openssl genrsa -des3 -passout pass:x -out server.pass.key 2048
  • openssl rsa -passin pass:x -in server.pass.key -out server.key
  • rm server.pass.key
  • openssl req -new -key server.key -out server.csr
  • openssl x509 -req -sha256 -days 365 -in server.csr -signkey server.key -out server.crt
  • cat server.crt server.key > server.pem
• Letsencrypt : TODO

TODO

• Put data writing in a queue
• Put alert sending in a queue

Characterisation of the population of beings of Riverworld.

Riverworld is a fictional planet and the setting for a series of sci-fi books written by Philip José Farmer.
Riverworld is an artificial environment where all humans (and pre-humans) ever born who died after reaching 5 years old are reconstructed.
Most of the resurrected awaken in a body equivalent to that of their 25 year old selves, in perfect health and free of any previous genetic or acquired defects.

A friend of mine made fun of the book concept and claimed half of the beings ressucitated would be prehistorical.

This project is proving him wrong.

Definitions

• Begining of mankind : Homo Erectus, -1500K, assuming all beings on Riverworld can walk
• End of mankind : 2016. In the book, all people die in 1983 after interacting with an alien civilisation.
• Child Mortality CM : death of infants and children under the age of 5
• Infant Mortality IM : death of infants and children under the age of 1
• Life Expectancy LE : average time a being is expected to live
• Life Adult Expectancy LAE : average time a being is expected to live if he reachs 5 years old

Datas

• Dataset compiled amongst considered sources :
  • until now, conservative estimations for pre modern times : ./data/population-min.csv
  • until now, optimistic estimations for pre modern times : ./data/population-max.csv
  • until now, average estimations for pre modern times : ./data/population-avg.csv
  • including 2070 projections ./data/population-future.csv
• The dataset consists of Point In Times (PIT)
• For each PIT, these metrics are available : year, beings count in millions, LE, LAE, CM and continental proportions
• Depending on sources, CM, IM, LE and/or LAE are provided or not. Some datas have been extrapolated. Underlying model : LE = CM * 5/2 + LAE * (1 - CM)
• Case of beings count in prehistorical times :
  • Beings counts estimation flucuates a lot. They can go as low as 1K individuals up to 100K
  • We have 3 milestones in our dataset : -1500K (lower paleolithic), -50K (higher paleolithic) and -10K (begining of history)
  • For -10K, the poulation count is within the magnitude of the millions according to most of the sources. We kept the McEvedy estimation of 4 millions.
  • For -50K, we kept the higher estimation of Jean-Pierre Bocquet-Appel from his study of upper paleolithical meta populations in Europe. He found 15K individuals in the Aurignacien (-30K). Wordlwide, we estimated, as a minimum, we timed this result by 5. As a maximum by 10.
  • For -1500K, we've assumed the population could not be higher than in -50K. We assumed the population at that time would be the population of -50K / 2.
• Case of LE et CM for prehistorical times :
  • As for beings counts, the LE estimations vary greatly
  • The Kaplan study suggest hunter gatherer modern societies tell us how prehistoric men lived and died. The study suggests the ALE is around 50 years and the CM around 0.5
  • For reference, the CM of 1900 is 0.4 and the Scheidel estimation of Classic Rome CM is 0.5
  • We cowardly derived the CM of pre Roman times to 0.6

Calculus

• Linearity in between PITs :
  • The underlying assumption is that the PIT metrics evolve linearly in between two PITs
  • This assumption can be considered true from -1500K to 1700, from 1700 to 1900, from 1900 to 1950 and from 1950 to today
  • The PIT resolution in the dataset is consistent with this observation
  • We then assume numerical midpoint integration is a reasonable estimation
• For the period PIT 1 => PIT 2
• AB yx = Amount of Beings for year x
• Elapsed Time ET = y2 - y1
• Average Amount of Beings for Period AABP = (AB y2 + AB y1) / 2
• LAE for Period LAEP = (LAE y1 + LAE y2) / 2
• Proportion of Adult Beings To Be for Period PABTBP = 1 - ((CM y1 + CM y2) / 2)
• Thus Amount of Beings who were Born for Period ABP = ET * AABP / LAEP
• Thus Amount of Adult Beings who were Born for Period AABP = ET * AABP * PABTBP / LEP
• Another calculus method is to use a simpe model of population growth :
  • AB y1 = Ce^(r * y1) and AB y2 = Ce^(r * y2)
  • By integration, ABP = ET * (AB y2 - AB y1) / (ln(AB y2) - ln(AB y1)) / LEP
  • This method proved to be very similar to the naïve numerical integration we selected (cf Report - Figure 6)

Results

• The last report : ./output/(min|max|future)/report.pdf
• The last plots : ./output

Sources

• https://en.wikipedia.org/wiki/Human_evolution
• https://ourworldindata.org/child-mortality
• https://ourworldindata.org/infant-mortality
• https://en.wikipedia.org/wiki/Lifeexpectancy#Variationover_time
• https://en.wikipedia.org/wiki/Worldpopulation#Pastpopulation
• https://ourworldindata.org/world-population-growth
• http://www.unm.edu/~hkaplan/KaplanHillLancasterHurtado2000LHEvolution.pdf
• https://en.wikipedia.org/wiki/Prehistoric_demography
• https://en.wikipedia.org/wiki/Worldpopulationestimates
• http://www.evolhum.cnrs.fr/bocquet/jas2005.pdf
• https://www.princeton.edu/~pswpc/pdfs/scheidel/040901.pdf
• https://scholarspace.manoa.hawaii.edu/bitstream/handle/10125/17288/AP-v47n2-190-209.pdf
• https://ourworldindata.org/world-population-growth
• http://www.math.hawaii.edu/~ramsey/People.html

Install

• pip install --upgrade pip
• pip install -r requirements.txt --user
• Fonts used in this project : ./resources/fonts
• PDF dependencies :
  • mac : brew cask install wkhtmltopdf
  • linux : apt-get install wkhtmltopdf

Run

• Configuration is loaded from config.ini
• python riverworld.py min|max|avg|future
• all at once : python riverworld.py min;python riverworld.py max;python riverworld.py avg;python riverworld.py future

Credits

• Author : Lucas Mouilleron, http://lucasmouilleron.com
• Thanks to : Jean-Benoît Bourron, Romain Charlassier