2015 – 2016 · Saint-Étienne, France · On-site
Production engineer (system monitoring and operations menu)
Eovi-mcd — A major French mutual insurance group
The workload dropped after the data migration, and monitoring of the Windows and Linux servers came down to operations: Centreon was put in place, along with a console ranking incidents by priority and an alerting setup. The scope arrived without instructions.
The role took shape in contact with the systems team. First relaying critical alerts; then, that team being saturated, handling some of them directly, consolidating procedures so the rest of operations could handle them unaided, and reporting back the alert correlations observed. Restarting web services from the command line, orderly shutdown and restart of services, extending saturated tablespaces, purging filesystems filled with logs: first-line system administration was absorbed almost entirely.
The transactional system's startup and shutdown scripts, old and unreliable in ways nobody could characterise, were rewritten and commented. With their reliability restored, those switchovers could finally be handed to the scheduler — something that had not been risked before.
The project closed with a single operations menu covering every step of the runbook. Flow reception and consolidation, until then launched transfer by transfer and then concatenated and moved by hand, became a single keystroke; so did generating the morning reports and the evening scheduling. Operations now fitted into one script, with the essentials automated and monitored.
Achievements
Making executions reliable
A given job's pre- and post-processing are now invoked systematically, whatever the trigger event. Durations are measured separately for pre-processing, processing and post-processing.
Checking that pre- and post-processing exist
Verified at placement time, rather than discovered when the batch failed during the night.
Server monitoring put in place
Centreon, an incident prioritisation console and an alerting setup, on a Windows and Linux scope that arrived without instructions.
First automation project of the post
Carried out under the production manager's oversight; the trust earned on this occasion opened the automation projects that followed.