XDCR Custom Conflict Resolution Function Execution Timeout Configuration

Answer to the Core Question

1. Maximum execution time to enforce: There is no built‑in limit in Couchbase XDCR; the system silently falls back to last‑write‑wins if a function runs longer than an internal window (typically a few hundred milliseconds). To avoid silent failures, you should enforce an application‑level timeout of 5 seconds for your custom function. This value is a pragmatic compromise: it is long enough for most business logic (e.g., merging JSON, applying business rules) yet short enough to keep the replication pipeline responsive.

2. Can a timeout be configured via REST or Admin Console without a cluster restart? No. Current Couchbase Server releases (7.x and 8.x) expose no REST endpoint or Admin Console setting that controls the conflict‑resolution window. Any attempt to add such a parameter would require modifying the server binary or configuration files, which in turn requires a rolling restart.

3. How to enforce the timeout while preserving LWW fallback for genuine conflicts? Implement the timeout inside the conflict‑resolution function itself. Wrap your logic in a Promise.race (or equivalent) that rejects after the chosen deadline. If the function times out, return null or throw an exception; XDCR will treat this as a failure and apply the default LWW rule. This preserves the existing fallback path while giving you a visible failure mode.

Practical Steps

1. Determine your Couchbase Server version:

   curl -s http://localhost:8091/pools | jq '.serverVersion'

2. Inspect the XDCR bucket configuration to confirm no timeout setting is present:

   curl -s http://localhost:8091/pools/default/buckets//xdcr | jq .

3. Modify your custom conflict‑resolution function (JavaScript, Go, or Java) to include a timeout. Example in JavaScript:

   function resolveConflict(doc, revs) {
     const timeoutMs = 5000;
     return Promise.race([
       actualResolveLogic(doc, revs),
       new Promise((_, reject) => setTimeout(() => reject(new Error('Timeout')), timeoutMs))
     ]);
   }

4. Deploy the updated function to the XDCR target bucket. No cluster restart is needed; the function is loaded on demand.
5. Enable XDCR debug logging to verify timeout behavior:

   curl -X POST http://localhost:8091/pools/default/logging -d "logLevel=xdcr:debug"

   Check logs for entries like Conflict resolution timed out or Last‑write‑wins applied.

Considerations and Caveats

• Setting the timeout too low may increase LWW fallbacks for legitimate, heavy‑weight resolutions.
• If your function must perform external network calls, consider retry logic that respects the same deadline.
• Monitor replication lag after deployment; a sudden increase may indicate too aggressive a timeout.

Missing Diagnostic Detail

To fine‑tune the timeout, I need the current Couchbase Server major version (e.g., 7.6, 8.1). This will confirm whether any undocumented flags exist and help set a realistic deadline.