2024-10-03 · Oliver Grant

Indexer Outages and How Teams Rehearse Reorgs

When an indexer falls behind, dashboards lie quietly. Users blame the frontend; engineers discover block lag hours later. Reorg handling is the unglamorous skill that separates hobby projects from operable systems.

In our Indexer & Data Pipeline Design course, week five simulates a shallow reorg on a forked network. Students must detect, replay, and verify row counts against a source-of-truth node. The exercise takes ninety minutes; preparation handouts take longer.

Teams that rehearse quarterly report calmer incident channels. We recommend documenting: detection signals, replay order, customer communication templates, and when to pause writes. None of this requires exotic tooling — cron plus structured logs often suffice early on.

If you maintain a subgraph today, compare its reorg behavior against a custom indexer for the same events. The comparison alone surfaces assumptions you may not have articulated.

#data#operations