Using detailed data on conflict-related incidents in Indonesia, we exploit seasonal variation in the relationship between rainfall and agricultural production to study the mechanism linking climate change and conflict. Furthermore, we ask whether irrigation and dam infrastructure help mitigate this link. We find that wet-season rainfall decreases production while rainfall during the dry season is beneficial for production. If agriculture is the mechanism through which climate change affects conflict, then we should expect the opposite effect on conflict, but with one-year lag. Our results show that, as expected, dry-season rainfall decreases conflict in Indonesia and in agricultural regions like Java, while wet-season rainfall increases conflict. In the latter, we find that irrigation increases conflict instead of reducing it. For Indonesia, irrigation reduces the effect of conflict during the dry season and amplifies it during the wet season. A plausible explanation is that the irrigation network is not well adapted to agriculture necessities which could generate civil unrest when a weather shock occurs. A policy that aim to reducing the impact of climate change on civil conflict should consider these drawbacks.