1 Reply Latest reply on May 6, 2013 6:28 PM by FreddieEssex

    FAL occasionally returns "CORRUPTION DETECTED"

      I have an 11g DataGuard implementation on Solaris that works most of the time. Occassionally, FAL will fail to transfer an archivelog file and report "CORRUPTION DETECTED". This is a red herring error message.

      I can scp the log file and apply it with RMAN - the log is not corrupted. There is a firewall on the standby machine which reports:

      *Built inbound TCP connection 203578275 for outside:192.x.x.x {aaaaaaaaaaaa} /36935 ... to inside:10.x.x.x {aaaaaaaaaam} /1521 ----- Today 12:48:48*

      *Teardown TCP connection 203578275 for outside:192.x.x.x {aaaaaaaaaaaa} /36935 ... to inside:10.x.x.x {aaaaaaaaaam} /152 duration 0:00:06 bytes 16819722 TCP Reset-I ----- Today 12:48:54*

      *Deny TCP (no connection) from 192.x.x.x {aaaaaaaaaaaa} /36935 ... to 10.x.x.x {aaaaaaaaaam} /1521 flags ACK on interface outside ---- Today 12:48:54*

      An rfs trace on the standby shows:

      Identified standby redo log 91 for implicit mid-log reconnect
      **** 2013-05-05 19:55:01.024 13841 krsr.c*
      RFS[17]: Header received for log 91 thread 1 sequence 10817 dbid 1311470586 branch 795203194

      Corrupt redo block 52075 detected: BAD CHECKSUM
      header size = 16, block size = 512
      Flag: 0x1 Format: 0x22 Block: 0x0000cb6b Seq: 0x00002a41 Beg: 0x0 Cks:0x75e0

      Sometimes after many tries some of these so-called "CORRUPTION DETECTED" logs actually make it through and get applied just fine. But sometimes, FAL just gives up and it never comes across the TNS stack.

      Any thoughts on what might be going on with the DG/TNS log transport?