Enterprise Replication

Define Server Fails

Although this is a warning it is actually an error, the cdr define fails, the time delta between the servers is too large

15:24:03  CDR NIF site: 1 <grp_student3> clocks differ by -77279 seconds
15:24:03  Warning - The operating system time of day clock at the
         remote site differs from the local site. This can
         slow down the data-sync apply performace at the
         target server. This happens only if timestamp or
         stored procedure conflict resolution rules are
         being used.
15:24:03  CDR NIF site: 1 <grp_student3> clocks differ by -77279 seconds

This is an easy one to miss especially if the machines are in different timezones

Logged and Unlogged Smart Spaces

Back in the day the smart spaces didn't need to be unlogged but that is no longer the case so how do you fix it. It's an easy fix

  • Create raw disk space and create soft link in the chunks directory.
  • Create the new smart blob space as logged
  • Shutdown the db engine and then bring back to quiescent mode
  • Ensure all ER transaction are cleared from the existing sbspace.
onstat –g ddr
onstat –g rqm BRIEF
onstat –g smb lod
  • Shutdown db engine and edit onconfig parameter to point to new sbspace
  • Restart db engine and ensure start-up is clean by monitoring logs on server and ER targets.

Dropping Deletion Tables

In the syscdr database, you need to examine the deltabdef table. The tabname, owner, and dbname are the names of the table that the delete table is a shadow of. The deltabid is the number portion of the delete tabe id. (the # part of cdr_deltab_######). The orphans can be identified by comparing the number part of cdr_del_###### with the deltabid within the deltabdef table. If the number is not found in the deltabdef column, then it is an orphan and can (or rather should) be dropped. N.B. If you manually delete the syscdr database, then you must manually drop the shadow delete tables as well. If this is not done, then you can run into a 57 error when trying to add replicates later on.

The preferred way to get rid of the syscdr DB is to run cdr remove (if ER is down) or cdr delete server (if ER is running).

High Availability Replication

High Availability Replication

Neat Trick

A neat trick to start HDR without backing up to somewhere real

ontape –s –L 0 –F –t STDIO | ssh <secondary> “. /etc/profile.d/ids_env.sh; ontape –p –t STDIO”

Mach 11


Personal Tools