I increased "ConnectionRetryInterval" time and also set "CheckRevision" to "true" but didn't help. I do not think it a concurrency problem. As mentioned
here, it is caused by continuing on a failed transaction.
I am using Postgres 14 and orthanc 1.11.3.
With housekeeping disabled, calling /tools/reconstruct ("ReconstructFiles": false) returns 200. But calling reconstruct with ("ReconstructFiles": true) fails with no error.
I removed database files and imported dicom files again on new instance of orthanc but error repeated.