|
we met a stranger question, we use the master/slave architecture to sync data.
the master server in Beijing, and slave server in ShangHai, some instances are success, and some instances always failed.
the feedback info is :
build index done. scanned 23741307 total records. 163.827 secs
Thu Jan 22 20:48:18.320 [replslave] Socket say send() errno:104 Connection reset by peer xxxx:27018
Thu Jan 22 20:48:18.320 [replslave] repl: DBException socket exception [SEND_ERROR] for xxxx:27018
Thu Jan 22 20:48:18.320 [replslave] Socket flush send() errno:104 Connection reset by peer xxxx:27018
Thu Jan 22 20:48:18.320 [replslave] caught exception (socket exception [SEND_ERROR] for xxxx:27018) in destructor (~PiggyBackData)
Thu Jan 22 20:48:18.320 [replslave] repl: sleep 2 sec before next pass
Thu Jan 22 20:48:20.321 [replslave] repl: syncing from host:100.80.1.30:27018
Thu Jan 22 20:48:20.427 [replslave] An earlier initial clone of 'feedback' did not complete, now resyncing.
Thu Jan 22 20:48:20.427 [replslave] resync: dropping database feedback
could you tell me what the reason is.
thanks for you reply.
Reporter: yuwen
E-mail: mjjjj2001@163.com
|