{"version":"2.1.309","anchor":"network-down-no-response-retries-get-a-separate-reconnect-bu","canonical_anchor":"network-down-no-response-retries-get-a-separate-reconnect-bu","heading":"Request retries get a separate reconnect budget for network drops","tier":"notice","area":"API Requests","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.309\/e\/network-down-no-response-retries-get-a-separate-reconnect-bu","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.309","markdown":"### Request retries get a separate reconnect budget for network drops\n\nWhen the network drops and no response arrives, Claude Code now retries from its own reconnect budget, limited by maxReconnects\n\n**What**\n\nWhen a request to the API fails, Claude Code keeps a budget of how many more times it may retry. This release adds a separate budget for reconnecting after the network goes down:\n\n- If a request gets no response and the failure is flagged `networkDown`, Claude Code retries as long as its reconnect count is below `maxReconnects`. Each such retry adds one to a new `reconnects` counter. Before, these failures went through the general failure handler.\n\n- The retry tracker now reports `reconnectsLeft`, calculated from `maxReconnects`. Before, it had only `retriesLeft` and `sleptThroughLeft`.\n\n**Why**\n\nA brief network drop now uses its own retries instead of the general ones. This should help requests survive dropped connections, while `maxReconnects` caps how many times Claude Code tries to reconnect.\n\n- Area: API Requests\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}