Abort: cannot lock repository when using hg serve on low‑traffic deployments

Goal

Reduce I/O and CPU costs for a Mercurial repository that receives very few requests per day.

Constraint

Mercurial protects its store with a lockfile located at .hg/store/lock. When hg serve spawns multiple worker threads, each thread may attempt to acquire this lock even for read‑only operations, potentially triggering the documented error abort: cannot lock repository.

Unresolved Decision

The official guides do not prescribe default values for --workers or --timeout in hg serve. Choosing the optimal thread count and request timeout for a low‑traffic scenario remains an open tuning question, directly affecting lock contention and cost.

Questions

• What minimal worker count for hg serve balances cost savings and avoids lock contention on a repository with fewer than one commit per day?
• Does the --timeout setting influence how often the cannot lock repository error occurs, and what default would be appropriate for low‑traffic use?
• Is there an official or community recommendation for disabling the lockfile in read‑only hg serve instances to reduce I/O overhead?