dragonflydb / dragonflydb/dragonfly
Document and expose WAN-oriented replication backlog sizing guidance
- Dominant language
- C++
- Stars
- 31.6k
- Forks
- 1.3k
- Avg merge
- 1d 10h
- Merged PRs (30d)
- 137
Description
## Problem
The replication backlog defaults to a five-second age target and 0.5% of `maxmemory`. Under
continuing writes, a WAN outage longer than this window can evict the required LSN and force a full
database synchronization.
The behavior is intentional, but the defaults target short connection restoration rather than
rack or DC partitions.
This behavior is documented, but its effect on WAN recovery still deserves a dedicated issue.
## References
https://github.com/dragonflydb/dragonfly/blob/980c70331d194d7133079917668fec269444c65c/src/server/journal/journal_slice.cc#L23-L49
https://github.com/dragonflydb/dragonfly/blob/980c70331d194d7133079917668fec269444c65c/src/server/journal/journal_slice.cc#L176-L207
https://github.com/dragonflydb/dragonfly/blob/980c70331d194d7133079917668fec269444c65c/src/server/dflycmd.cc#L486-L495
https://github.com/dragonflydb/dragonfly/issues/7994
Contributor guide
Research direction
Read the replication backlog defaults in src/server/journal/journal_slice.cc at the referenced sections, then inspect the related handling in src/server/dflycmd.cc and issue #7994. Document how the five-second and maxmemory-based defaults affect WAN recovery, and expose clear sizing guidance for that scenario.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- databases, distributed-systems
- Issue type
- Documentation
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 68/100