Borys

We need to refactor RegisterOnChange in a way that removes mutex lock via the LockChangeCb and UnlockChangeCb methods from the Transaction code

Today we check the migration only by comparing node id. In case migration slot range was updated in config we need to cancel the migration and start a new migration...

TTL exhausted error during cluster migration

1

We have a deadlock in our cluster code. lock config_update_mu_ mutex interferes vs dispatchTracker (we can not use it during PAUSE) 1) we try to set a new config via...

CmdArgParser::ToUpper needs to be removed

1

CmdArgParser::ToUpper function is useless in most cases because command comparison should be done case insensitive (we need to check this and fix if this isn't done). Also this function semantically...

Data loss during heartbeat

2

When we do snapshots (replica/migration) we use global transactions to prevent simultaneous data changes during db_slice and journal_slice RegisterOnChange function calls to prevent data loss, but heartbeat ignores global transactions...

https://github.com/dragonflydb/dragonfly/actions/runs/10205678288/job/28236983140

1. Displacement improving: Instead of displacing (2 bits that describe left and right alternative cell), we can use shift (2 current bits can provide us with the next 3 positions...

given: two migration processes from one node with a replica to another 2 nodes the size of the data 2GB migration for all slots before migration source node memory consumption...

fixes: #3870 Before BM_Grow 721975 ns 721914 ns 1020 Now BM_Grow 598066 ns 597285 ns 1237