Bugs found for distributed systems

Here I list some bugs found by my fuzzing tools for distributed systems and databases.

• Braft/brpc
  • memory leak
  • Fail to rename
  • No exception handling after error?
  • Check failed: meta.term == header.term
  • data loss after restart?
  • memory leak if binding failed
  • Node cannot response after one node stepped down
  • TaskControl may interrupt non-exist threads
• C-Raft
  • ASan reported heap-buffer-overflow
  • ASan report bad-free
  • ASan reported SEGV
  • example/server failed to start after killing
  • Fix memory leak when crc check failed.
• NuRaft
  • Got FATL logs when test with the calc example
  • Exit if ctrl-d received
• RedisRaft
  • Fix heap buffer overflow
  • Assertion `cache->start_idx + cache->len == idx’ failed
  • Radis raft panic
  • Assertion failed.
• RethinkDB
  • Memory bugs reported by ASan
  • Guarantee failed: [iterator_and_is_new.second] value to be inserted already exists.
• Aerospike
  • stack-buffer-underflow
  • error creating fabric published endpoint list
  • dirty read if the connection is not stable?
• ClickHouse
  • Distributed table cannot be used after rename column
  • Distributed table cannot find column with “greater” query
  • Not found column in substr
  • Constraint check throws Missing columns exception
  • Does distributed table support select with order?
  • Always use default database when creating distributed table
  • Cannot drop table after a wrong create distributed table statement executed
  • Cannot create table after drop
  • Create Distributed table can succeed if the database not exists
  • Fatal error when createDatabase
  • Got weird fatal log: floating point inexact result
  • Logical error: ‘It’s new replica, but database is not empty’
• etcd
  • runtime error: slice bounds out of range
• ZooKeeper
  • Committing zxid 0x100000003 but next pending txn 0x100000002
  • NullPointerException in SendAckRequestProcessor
  • Zookeeper crashes after commit fail