Borys

different status for keys number in "in" and "out" migrations

2

In the test_cluster_fuzzymigration when migration is finished I see different numbers of keys for "in" and "out" migrations.

fix: reduce delay when stop replica #3020

1

this should fix the problem: Running replicaof no one on replica in full sync state returns after more than 20 seconds the problem was that after we shutdown the socket...

``` libc.so.6!__pthread_kill_implementation(int no_tid, int signo, pthread_t threadid) (pthread_kill.c:44) libc.so.6!__pthread_kill_internal(int signo, pthread_t threadid) (pthread_kill.c:78) libc.so.6!__GI___pthread_kill(pthread_t threadid, int signo) (pthread_kill.c:89) libc.so.6!__GI_raise(int sig) (raise.c:26) libc.so.6!__GI_abort() (abort.c:79) google::DumpStackTraceAndExit() (/home/borys/projects/dragonfly/build-dbg/_deps/glog-src/src/utilities.cc:178) google::LogMessage::Fail() (/home/borys/projects/dragonfly/build-dbg/_deps/glog-src/src/logging.cc:1949) google::LogMessage::SendToLog(google::LogMessage * const...

feat: add reusable fibers and fiber pool

2

It's just an idea. On my PC for test: memtier_benchmark -t 8 --pipeline=30 --hide-histogram --test-time=3000 --distinct-client-seed -d 256 --ratio=1:0 I've got 19% performance improvement

JournalStreamer::WriterFb is a fiber that exists only because it uses a synchronous socket API but the socket also implements AsyncSink API so it is possible to get rid of this...

fix for #3227 during migration finalization, we didn't check the MIGRATION CONFIG and used the base one.