We see this error on a number of tables when doing an expdp on 12c / Windows.

A restart of the Oracle database service makes this go away...

Can anyone shed some light on what may cause this and the best way to manage it.


[oracle@orcluat ~]$ oerr ora 01555
01555, 00000, "snapshot too old: rollback segment number %s with name \"%s\" too small"
// *Cause: rollback records needed by a reader for consistent read are
//         overwritten by other writers
// *Action: If in Automatic Undo Management mode, increase undo_retention
//          setting. Otherwise, use larger rollback segments

During data pump export, it's session reads rows from tables being exported and rows are consistent to the point in time when the expdp session has started reading the table. If you want all tables in the export to be consistent to the same point in time, you need to use the FLASHBACK_SCN or FLASHBACK_TIME parameter.

If the rows of the table being read are changed then server process has to make the rows consistent by using undo information. But it is unable to obtain the required undo, which might have been replaced by new undo information(controlled by undo_retention,there are certain condition where undo retentation may be ignored, initialization parameter).

Generic Solutions:

As we don't know what exactly had happened in your case, we can only give some generic solutions.

  1. As per the oerr utility, increase the value of undo_retention, presumably you are using automatic undo management. Make sure you have enough space on undo tablespace to satisfy the retention time.
  2. Try to avoid business hour while initiating export job.
  • the server process reads the undo information, it does not apply it – miracle173 May 9 '17 at 5:54
  • Thanks so much for the clear and concise response. My apologies for the briefness of my question. We think we have identified an external data warehousing process which is performing transactions on particular tables which we think could lead to this condition. We are thinking we could export the schema into a new schema specifically for this process so normal operations would not be affected. – dank May 9 '17 at 17:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.