10 Replies Latest reply on Mar 30, 2017 8:49 AM by Toby Erkson

    Historical Failure.  What the...?

    Toby Erkson

      Was the error so epic that it made history in the Tableau Server's mind?

        • 1. Re: Historical Failure.  What the...?
          Matt Coles

          We got tired of hearing "Epic Fail!" and substituted this instead.


          No, really though--that means the number of failures exceeded the concurrent failure threshold.

          1 of 1 people found this helpful
          • 2. Re: Historical Failure.  What the...?
            Toby Erkson

            Thanks Matt but I don't quite understand.  "Concurrent failure threshold", does that mean that there were a number of errors during the operation so they all just got lumped into this single error message (an error message for too many error messages )?

            Matt Coles wrote:


            We got tired of hearing "Epic Fail!" and substituted this instead.


            Actually, that would be a pretty funny error message

            • 3. Re: Historical Failure.  What the...?
              Matt Coles

              Let's say you've got a daily extract schedule on your 10.0 Server. You publish a workbook to it and everything is fine. But oh no! Your DBA accidentally removes your user account from the DB server it was extracting from! If it fails N times in a row, the extract refresh is automatically disabled and will no longer be attempted. However, it still gets a record dumped every day, on the exact same schedule, into the background tasks table in the Postgres repo DB, with the Notes field being set to "Historical Failure". I'm not sure why it keeps putting records in there--I assume that the design is to provide visibility to the Admin that there's still something broken that needs attention.

              4 of 4 people found this helpful
              • 5. Re: Historical Failure.  What the...?
                Obed Tsimi

                Very Helpful!

                • 6. Re: Historical Failure.  What the...?
                  Jeff Strauss

                  thanks for the details, I'm sure many of us would have had that look of confusion at some point from seeing this message.

                  • 7. Re: Historical Failure.  What the...?
                    Rick Kunkel

                    For what it's worth, this is in the documentation but is a little hard to find because the phrase "historical failure" do not appear there.  I know because I ran into this exact same problem and had trouble finding an answer.




                    Error wordage for future versions has been changed to, if I recall correctly, "Suspended: skipped".  I don't know if that will appear in 10.0.x though.

                    • 8. Re: Historical Failure.  What the...?
                      Mark Wu

                      This new 'Historical Failure' confused me a lot when I first saw it specially from Day 1 when server was upgraded to v10. After hours of research, it starts to make sense for me. I wanted to share my findings here.


                      1. Conceptually, when an extract failed a few times in a row, specially sometimes it failed after backgrounder run many minutes, why you want to waste backgrounder knowing very likely that the extract job w fail again? I think that this is why v10 came w this new feature : if the job failed a few times, do not run it anymore.


                      2. Now what error should Tableau provides in the above scenario? It is not right for Tableau to mention the same error as happened before, so Tableau designed a new error type - 'Historical Failure'.


                      3. How many failures in the row before 'Historical Failure' error? For a hourly refresh, I counted 5 failures in a row before 'Historical Failure'. I am not sure if this is true for daily or other schedule.


                      4. How do you really know what was the problem when you saw 'Historical Failure'? You really do not know what is the problem. You will have to give specific extract name in the admin view 'Background Tasks for Extracts'  tab, then look for the first a few failures before 'Historical Failure to find out what was the true error msg.


                      5. The first a few failures before 'Historical Failure' occurred will show you the details why the extract actually failed. You will need this true error message for you to fix the problem.


                      6. Why I saw some 'Historical Failure' error on Day 1 after v10 upgrade? I have some hourly jobs failed. When I noticed the extract failure, I normally check the most recent ones. For this case, it gave me mostly 'Historical Failure' errors, which confused me a lot. I had to dig into the early failure and read this post before I figured out what happened.

                      4 of 4 people found this helpful
                      • 9. Re: Historical Failure.  What the...?
                        Matt Coles

                        Mark, you've got it correct. The number of consecutive failures allowed before a schedule for a given workbook or data source is suspended is configurable, but only at the Server level. You can configure the number with a tabadmin set command:


                        tabadmin set backgrounder.failure_threshold_for_run_prevention 10
                        tabadmin configure
                        tabadmin restart


                        ...would increase the failure threshold from the default of 5 consecutive failures, to 10.


                        Note that the messaging has now changed, thankfully, and instead of "Historical Failure", you get "Suspended: skipped"

                        • 10. Re: Historical Failure.  What the...?
                          Toby Erkson

                          You can still get the warning messages without turning off the scheduled extracts:  Re: Adjusting refresh failure email settings possible?

                          This is something I elected to do.