witt_reports

Witt Reports
High level Description
Links
Operations
Known Issues
Super Special Knowledge
Experiments / Results / Knowledge Base
Dependencies
Incident Log
Authors

Witt Reports

This project was initiated by a customer request: https://trello.com/c/o1Rs0TcH/259-witt-api-for-sold-out-products-in-visualytics The reports need to be updated daily (except style_cover, which changes weekly)

High level Description

This solution is split into two components:

A FastAPI service that provides two endpoints to generate the required CSV files
A bash script that calls the service and stores the result in a location that can be accessed by the customer<HTML></ol></HTML>

Operations

The service to generate the report lives on pci01.picalike.corpex-kunden.de, while the cronjob to store the report on a webserver lives on frontend01.picalike.corpex-kunden.de

Deploy this service

It is installed via docker. In the git are bash skripts for deployment to pci01. It follows the typical

bash ice_tea_build.sh
bash ice_tea_push.sh
bash ice_tea_deploy.sh

cycle.

Start this service

on pci01: cd ~/docker_bin/witt_reports/ ; bash ice_tea_run.sh
on frontend01: bash ~/bin/witt_reports.sh

Use this service

curl http://pci01.picalike.corpex-kunden.de:8320/style_coverage > /tmp/style_cover.csv # should be fast
curl http://pci01.picalike.corpex-kunden.de:8320/availability > /tmp/avail.csv # could take a while

Repair this service

your guess is as good as mine
restarting the docker container on pci01 might help

Known Issues

not connected to any serivce discovery service (intentionally not connected to zookeeper)
logging does not have timestamps
logging needs general improvement

Super Special Knowledge

The webservice performs a lot of calls to pci-server:/get_product_metaData in an asynchronous fashion.
The list of witt shops is currently hard coded in main.py (problem for future self)
pci-server is hard coded to http://pci01.picalike.corpex-kunden.de:8090
the docker container is restarted every morning at 7 AM to avoid memory issues
the report generation is triggered via cronjob on picalike@frontend01.picalike.corpex-kunden.de (not HPC)

Experiments / Results / Knowledge Base

You performed a test? share it.

Dependencies

PCI: http://dokuwiki.picalike.corpex-kunden.de/pci_doku

Incident Log

Authors

Johannes Otto j.otto@picalike.com
Bjoern Zessack b.zessack@picalike.com<HTML></ol></HTML>

Table of Contents