Berkeley DB Replica Mechanism-master-Slave synchronization

REPMGR/REPMGR_NET.C,

__repmgr_send (): Do Send_broadcast, then deal with the db_rep_permanent according to policy

__repmgr_send_broadcast (): On each site, send_connection ().

MASTER Send

Log/log_put.c, Log_put (),

Do not accept Rep_client

__rep_send_message (env, Db_eid_broadcast,-Rep_newfile, Rep_log

Txn/txn_chkpt.c, __txn_checkpoint ()

Rep_client is here only when recover, sync MP quits immediately

Before Master Sync MP, send to client:

__rep_send_message (env, Db_eid_broadcast, Rep_start_sync

Waiting for Chkpt_delay

Write CKP Log Rec

Db_rep_rerequest vs Db_rep_anywhere

* GAP Requests is "new" and can go anywhere, unless
* This is already a re-request.

Repmgr_net.c, __repmgr_send ()

if ((Flags & Db_rep_anywhere | db_rep_rerequest)) ==db_rep_anywhere &&

(site = __repmgr_find_available_peer (env))

after sending, db_rep_permanent, check policy, how many ACK is required to return (for durability).

Http://docs.oracle.com/cd/E17076_03/html/api_reference/C/repmgrset_ack_policy.html

Default Db_repmgr_acks_quorum, Repmgr_net.c, __repmgr_send (), Visible (n-1)/2 + 1

After determined, __repmgr_await_cond (env, got_acks, &perm, Rep->ack_timeout, &db_rep->ack_waiters) ;

= = while (Got_acks (env, &perm)) {pthread_cond_timedwait (&db_rep->ack_waiters,rep->ack_ Timeout)}

Repmgr_net.c, Got_acks ().

Dbinc/rep.h,

struct __db_rep {}

Repmgr_runnable *selector, **messengers, **elect_threads;

WSAEventSelect

Client Receive

REPMGR_METHOD.C, __repmgr_start_int ()-Elect/msg/select threads
REPMGR_METHOD.C, __repmgr_start_selector ()
REPMGR_SEL.C, __repmgr_select_thread ()
REPMGR_WINDOWS.C, __repmgr_select_loop ()
REPMGR_WINDOWS.C, Handle_completion ()-
REPMGR_SEL.C, __repmgr_read_from_site ()
REPMGR_SEL.C, Dispatch_msgin ()-

Put Db_rep->input_queue, __repmgr_signal (&db_rep->msg_avail)

Rep.h, struct __db_rep- cond_var_t check_election, Gmdb_idle, msg_avail;

REPMGR_METHOD.C, __repmgr_start_int ()
REPMGR_METHOD.C, __repmgr_start_msg_threads ()
REPMGR_MSG.C, __repmgr_msg_thread ()
Message_loop ()

while (ret = __repmgr_queue_get () ...

__repmgr_queue_get-while (M = available_work (env)) = = NULL), wait on Msg_avail
Process_message ()
REPMGR_RECORD.C, __rep_process_message_int ()

For Rep_log messages, call

REP_LOG.C, __rep_log ()

REP_RECORD.C, __rep_apply ():

Log.h, struct Log {}-Waiting_lsn, Max_wait_lsn, __db.rep.db, READY_LSN

Waiting_lsn:it is the first LSN that we are holding without putting in the log, because we received one or more log rec Ords out of order.

Ready_lsn:it is the next LSN of we expect to receive. It s normally equal to "LSN", except at the beginning of a log file, at which point it's set to the lsn of th E first record of the new file

If it is the next log rec that we need, call __rep_process_rec (), __rep_remfirst/__rep_getnext then processes the log in TMP db; __rep _loggap_req ().

If it is behind the log rec we need, enter TMP db, update WAITING_LSN, send __rep_loggap_req ().

If you are in front of the log rec we need, we will receive duplicate log rec.

REP_RECORD.C, __rep_process_rec (), NewFile Special handling exit. In addition to CKP, the other writes directly to the log file first

-Db___txn_prepare: Flush log directly

-Db___txn_regop: __rep_process_txn (), get the required write lock, get the TXN corresponding to all the log rec, sort, read out, Db_dispatch (db_txn_apply

-DB___TXN_CKP: First write a rec in rep_db (bookkeeping DB), nooverwrite, if there is already, then other threads are doing CKP, exiting. Sync MP; Write DB_LOG_CHKPNT log rec, flush log

Berkeley DB Replica Mechanism-master-Slave synchronization