Purge Caches corrupting Xlates

[Kathy Zwilling]

We are on 5.2 rev 1 on Unix HP 11i.  

I have seen a few posts talking about this subject with 5.x and up and am wondering if we are experiencing this issue.  

After a purge caches is done, we may have a single message fail and then the following ones are fine.   I can run that failed message thru the testing tool and it processes just fine.   I can send it out the outbound interface and it processes fine in the receiving system.

I haven’t been able to put a finger on what is causing the problem because the message in the engine logs haven’t made any sense.  However, I wonder if the purging is corrupting the translation in memory or maybe the HL7 format file in memory.  I have looked at the failed message and it doesn’t contain any of the segments specified in the engine log error messages.   Any help would be greatly appreciated!

The messages in the logs look like this:

Engine idle — 04/25/2005 15:53:24

04/25/2005 15:53:29

[msg :RecD:ERR /0:results01_xlate] Unable to load segment ‘AL1’

04/25/2005 15:53:29

[msg :RecP:ERR /0:results01_xlate] [0.0.6702604] ERROR.  Invalid message type ‘ORM^O01’.

04/25/2005 15:53:29

[xlt :xlat:ERR /0:results01_xlate] [0.0.6702604] Xlate ‘cita_care_ORMO01.xlt’ failed: Input validation errors: unable to get information on invalid HL7 message ‘ORM^O01’ for parse

04/25/2005 15:53:30

[msg :RecD:ERR /0:results01_xlate] Unable to load segment ‘BLG’

04/25/2005 15:53:30

[msg :RecP:ERR /0:results01_xlate] [0.0.6702611] ERROR.  Invalid message type ‘ORM^O01’.

04/25/2005 15:53:30

[xlt :xlat:ERR /0:results01_xlate] [0.0.6702611] Xlate ‘cita_care_ORMO01.xlt’ failed: Input validation errors: unable to get information on invalid HL7 message ‘ORM^O01’ for parse

Engine idle — 04/25/2005 15:54:10

Topic

[Ryan Boone]

Kathy,

We used to purge caches after table updates all the time — until we discovered afterward that segments inside of xlate iterates got messed up. However, I don’t believe we had the problem with just one message — it would be any message after the purge. When upgrading from 3.5.2 to Platform 5.3, I verified that the problem still occurred, so we simply cycle the processes to avoid the problem. (We haven’t installed any revisions yet, though.)

[Kathy Zwilling]

Thanks for your response.   Just to clarify…when you cycle the process you are saying to shut it down and back on…correct?

[garry r fisher]

Hi Kathy,

We thought we had this problem with v5.3 and were advised to updated to rev2 as there was a specific fix for this problem in that patch release.

As it happened we didn’t have this problem but hopefully this may help as it is specifically mentioned in the release notes for 5.3 rev2.

Regards

Garry

[Ryan Boone]

Yes, we cycle (shut down and restart) the entire process(es) when we update tables. It doesn’t take much longer, but you want to be careful not to “rush” the engine if you’re cycling more than one process. Just cycle them one at a time and let everything shut down and restart before you issue additional commands.

As Gary said, I seem to remember that there was a fix for the purge caches in the next rev (if you’re on 5.3, which you are not), but I don’t remember if it corrected the problem with xlate iteration segments or not.

[Rob Abbott]

The purgex corruption was fixed in 5.3 rev2.  I do not believe the fix is available in any other release.

Rob Abbott
Cloverleaf Emeritus

[Author]

Replies