• S-ASH Simulator
         sashpack.sh Nov 15, 07 .
    • changed all objects to be "sash_" instead of "v$ash_"
    • lots of changes, sort of new start
         ashpack.sh Oct 23, 07 .
    • added update to SQL collection
      collect last_found date for SQL
      collect found_count for # of times SQL found
      still don't have SQL purge routine but should be easy
      using deleate where last_found < some date
         ashpack.sh Sept 28, 07 .
    • added full field list to inserts to address "not enough fields"
         ashpack.sh Aug 20, 07 .
    • changed partition creation code variable from 1000 to 4000
      had error where size was exceded and partition creation failed
         ashpack.sh July 30, 07 .
    • changed for 10g to skip v$session_wait and just use v$session
    • for 10g, filter idle waits with v$event_name wait_class
         ashpack.sh June 27, 07 .
    • added init.ora parameter collection
    • added update of last collection date for SQL text to be used for purging
         ashpack.sh May 30 07 .
         ashpack.sh May 8 07 .

To do:

  • sql size & cleanout stuff
  • have an SQL_ID table to collect the sql_id or hash or addresss for the SQL to collect text and stats for instead of using the sqltxt table
  • field in v$ash_dbs that has last collect date or something
    to know if the target is acutally sendng data
  • put "poor mans" partitions back in for standard edition oracles
  • add wait class column to v$ash_all
  • fix up (fill in) wait class column on REPO
  • on REPO scan collected data with heuristics and fill a table with semi fast data collect list - table also meeds to be cleaned out to end semi-fast collectins
  • have a v$ table option (instead of x$) for system access only accounts (non sys)
  • Collect user names
  • Add purge routine for old SQL text and stats
  • Check on instance id field for RAC systems
  • Add a counter to SQL collection for # of times found
  • Track changes in init.ora (spfile)
  • Collect latch names
  • Add stay alive pinging and alive monitor job
  • Insert into v$ash a line even when no active sessions found in order to track sample_id for data mining queries
  • Collect sort info
         sorting info
  • Collect parallel query info
         pqo info
  • Add Keep column to data, to keep key areas beyond purge
  • possibly use
    to automatically reload new code on targets
  •      ashstart.sql
    to start collection on targets and clean up any old collections
  •      ashnow.sql
    consider using v$ tables instead of x$
    the only difference is gathering event names vs event ids
    I wonder if the over head is negligible
    using v$ would all system to run gather verses just sys
  • these DBIDs are a pain
    I'd rather use host.sid - it would be more readible
    plus I have problems with cloning and dbs being recreated
    on the same node with same sid, but getting difffernt DBID