2 Replies Latest reply on Feb 21, 2013 2:57 AM by ursusca

    "gc cr/current block lost" is  interconnector-related issue???

      Hi, all.

      The db is 2-node RAC ( on a linux machine.

      From time to time, I could see a lot of "gc cr/current block lost" and active sessions simultaneously.
      in normal time, the number of avgerage active sessions is around 50 , but at the time of system hang,
      the number has increased to 1000 that were waiting for "global cache" related event.
      Top 5 Timed Foreground Events
      Event Waits Time(s) Avg wait (ms) % DB time Wait Class 
      gc buffer busy acquire 184,701 74,764 405 29.25 Cluster 
      gc cr grant 2-way 907,467 36,359 40 14.22 Cluster 
      enq: TX - index contention 17,608 25,493 1448 9.97 Concurrency 
      gc current grant busy 96,833 23,778 246 9.30 Cluster 
      db file sequential read 2,304,140 13,442 6 5.26 User I/O 
      My questions are :

      1. "gc cr/current block lost" is caused by interconnector configuration or hardware issue??

      2. Otherwise, can "gc cr/current block lost" be caused by hot block(application issue)??
      The following senario is my guess.

      => node-1 requests a block for current mode, but node-2 is holding the block at current mode.
      => node-1 is waiting for the block to be released, but node-2 session does not issue "commit".
      => node-1 is waiting for a time period (30 seconds or 1 minute)
      => node-1 gives up the block and oracle increase the count of "gc block lost"
      => node-1 retries the block.

      3. if the problem was caused by interconnector issue,
      what could be the possible methods to track the history of interconnector performance at SYSTEM level.

      Thanks in advance.
      Best Regards.

      Edited by: 869578 on 2013. 2. 20 오전 12:50