Change Details

We have a cron job for the automatic banner update on wikipedia.de that is supposed to pull the newest version of the `wikipedia.de-banners` repository once a minute, to pull the latest banners and campaign data. Apparently, it does not do that. Suggestion: Instead of having individual commands, create a bash script in the home directory of the deploy user. Thse script should redirect all output to a log file (that is rotated with a logrotate configuration) and run with the [[ https://www.gnu.org/software/bash/manual/html_node/The-Set-Builtin.html | `-e` option ]] to stop if any command fails (alternative to using the `&&` operator).

We have a cron job for the automatic banner update on wikipedia.de that is supposed to pull the newest version of the `wikipedia.de-banners` repository once a minute, to pull the latest banners and campaign data. Apparently, it does not do that. Suggestion: Gabriel created a bash script in /home/deploy/bin/update_banners.sh that runs the cron job and writes a log file. But that script is written "in production" and should be formalized and improved upon in the infrastructure repo: * Install the script via Ansible (with the right permissions) * Change the `deploy` user crontab to use that script * Add a `MAILTO=fundraising-tech@wikimedia.de` to the `deploy` user crontab (see other crontabs in the playbooks on how to do it) * Add a logrotate configuration that purges/compresses the log file `/home/deploy/fetch-banners.log` every day and keeps a history of 7 days.

We have a cron job for the automatic banner update on wikipedia.de that is supposed to pull the newest version of the `wikipedia.de-banners` repository once a minute, to pull the latest banners and campaign data. Apparently, it does not do that. Suggestion: Instead of having individual commands,Gabriel created a bash script in /home/deploy/bin/update_banners.sh that runs the cron job and writes a log file. create a bash script in the home directory of the deploy user. Thse script should redirect all output to a log file (that is rotated with a logrotate configuration) and run with the [[ https://www.gnu.org/software/bash/manual/html_node/The-Set-Builtin.html | `-e` option ]] to stop if any command fails (alternative to using the `&&` operator). But that script is written "in production" and should be formalized and improved upon in the infrastructure repo: * Install the script via Ansible (with the right permissions) * Change the `deploy` user crontab to use that script * Add a `MAILTO=fundraising-tech@wikimedia.de` to the `deploy` user crontab (see other crontabs in the playbooks on how to do it) * Add a logrotate configuration that purges/compresses the log file `/home/deploy/fetch-banners.log` every day and keeps a history of 7 days.