On 11/13/19 4:40 PM, Brandon Ragland wrote:
Hello,
I have a Talend enterprise job that loads data into a PostgreSQL
database via the COPY command. When migrating to a new server this
command fails with the following error message:
org.postgresql.util.PSQLException:ERROR: COPY escape must be a single
one-byte character
Does the Postgres log offer anymore information?
Can you crank up the error level to get more info?
Are the encodings for the Postgres server/OS different ?
The thing is, I looked over the documentation for both Postgres 9.0 and
9.2. Both documentations say that multi-byte delimiters are not allowed.
So I'm very confused on why this job works perfectly on Postgres 9.0 but
not on 9.2.
I am unable to edit this Talend job, as it's very old and we do not have
the source code for the job anymore. I am unable to see what the actual
delimiter is. I am also unable to see exactly how the COPY command is
being run, such as whether it's pushing directly to the server via the
Postgres driver, or if it's created a temporary CSV file somewhere and
then loading the data into the server. I believe the reason we have
multi byte delimiters setup is due to the use of various special
characters in a few of the columns for multiple tables.
I am not aware of any edits to the source code of the old 9.0 Postgres
server.
The reason we are migrating servers is due to the end of life for CentOS
5. The new server runs CentOS 7. I believe that both servers are using
the default Postgres versions that come in the default CentOS
repositories. I know for sure that the CentOS 7 server is indeed running
the default Postgres version, as I installed it myself through yum.
Any help would be greatly appreciated.
Also, is there a way to copy the old Postgres server, dependencies, and
executables to our new server, in case the source was modified?
Brandon Ragland
Software Engineer
BREAKFRONT SOFTWARE
Office: 704.688.4085 | Mobile: 240.608.9701 | Fax: 704.973.0607
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
