Editing Scanco Administration (section)

====Error: Writing to UPAT database fails====

Sometimes when performing a morphological evaluation we get an error when writing to the UPAT database. This is entirely preventable. The error may look like this snippet from the LOG file:

<pre>
...snip...
Reading AIM file 6
DK0:[XTREMECT2.DATA.00002783.00016788]D0014390_TRAB_1N.AIM...
!> Reading AimVersion020
!> Image size after D1charCmp (run-length) uncompression will be:
!> Type of data               Char       1 byte/voxel
!> Total memory size          82.7       Mbyte
Trabecular Analysis (5)
starting calculations
Scanner-ID: 3401
Updating old record 16788
getting new eval record...
New Eval record:       58416
getting new slice records...
TIME:  8-MAR-2024 10:41:00.63
ERROR:
Writing indexed record (upat slice)
MESSAGE: %FOR-W-INCKEYCHG, inconsistent key change or duplicate key!
STATUS:            2
CODE:           50
FACILITY: FORTRAN
--------------
UPAT_FILL_DB V2.0-1
   Eval-Type: [5]:    Group-Code: [0]: Evaluation Type 6 Cortical
Analysis
    ISQ-File: [isqfile.ISQ]:      Aim-File: [aimfile1.AIM]:
Ct.Mask-File: [aimfile7.AIM]:       Ct-File: [aimfile8.AIM]:
...snip...
</pre>

The problem is because when running several jobs at once on the cluster there could be parallel attempts to write to the database. The solution is to ensure that all jobs that write 
to the database are run on the exact same queue. However, to leverage the cluster performance and running parallel jobs, we can split a typical morphological evaluation into two parts.
The first part is spread across all available queues, and the second part directs the final writing to database to a single queue (e.g., <code>SCEVAL_FAST</code>).

In <code>UCT_EVALUATION</code> or any <code>HIJACK</code>scripts, we need to replace:

<pre>
$ PAR1 = "EVAL:UE:UCT_EVALUATION_V6.COM|UE:IPLV6_UPAT_REDO_E5E6_XT2.COM|UE:UCT_EVALUATION_V6_UPAT_FILL_E5E6.COM|UE:UCT_EVALUATION_V6_CREATE_TIF.COM|UE:UCT_EVALUATION_V6_UPAT_PRINT_E5E6_NM.COM|"
</pre>

with 
<pre>
$ PAR1 = "EVAL:UE:UCT_EVALUATION_V6.COM|UE:IPLV6_UPAT_REDO_E5E6_XT2.COM|UE:STARTER_UPAT_POSTPROCESS_NM.COM|"
</pre>

The script <code>UE:STARTER_UPAT_POSTPROCESS.COM</code>, or in this case the version that does no matching (<code>UE:STARTER_UPAT_POSTPROCESS_NM.COM</code>) replaces the following three scripts:

 UE:UCT_EVALUATION_V6_UPAT_FILL_E5E6.COM
 UE:UCT_EVALUATION_V6_CREATE_TIF.COM
 UE:UCT_EVALUATION_V6_UPAT_PRINT_E5E6_NM.COM

We still can call <code>UE:IPLV6_UPAT_REDO_E5E6_XT2.COM</code> if we need to re-run the morphological analysis, but the other three scripts are specifically sent to <code>SCEVAL_FAST</code> to avoid any conflicts writing to the database.

If this doesn't fix the problem then also double-check that the sample/measurement number of the scan embedded in the ISQ is correct. There are occasions when that's been mistaken. See other FAQ entry about adjusting file header

[[Image:Redirect.jpg|15px]] [[#top|''Return to Contents'']] or [[Main_Page|''Return to Main Page'']]