dragonflydb / dragonflydb/dragonfly
Cluster fix READWRITE command
- Dominant language
- C++
- Stars
- 31.5k
- Forks
- 1.3k
- Avg merge
- 1d 10h
- Merged PRs (30d)
- 137
Description
Problem: The replica knows nothing about the cluster and always works in read-only mode. It means we can read from the replica all the data but don't get a -MOVE error if we try to write something.
Expected behavior: In READONLY mode we return -MOVE if we don't have the key or for any WRITE command. In READWRITE mode we return -MOVE error for any READ/WRITE commands
If we want to support this we will need to save in connection cntx a boolean set it if readonly command was invaded, and on cluster mode before invoking transaction read commands we need to check if this flag was set.
Where to start:
Service::VerifyCommandState() and IsUserAllowedToInvokeCommandGeneric
add
```
if (!IsClusterEmulated()) {
return cmd_cntx.rb->SendError(kClusterDisabled);
```
Contributor guide
Research direction
Start with Service::VerifyCommandState() and IsUserAllowedToInvokeCommandGeneric, then trace the connection context and cluster-mode command checks. Confirm the READONLY and READWRITE behaviors described in the issue, including the -MOVE responses, and verify the cluster-disabled path shown in the issue.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- databases, distributed-systems
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 32/100