COBOL - File Status 37 when opening sequential dataset with relative & dynamic options - file

Question

I'm trying to open a sequential dataset.
I get the file status 37, for which IBM says:
An OPEN statement was attempted on a file that would not support the
open mode specified in the OPEN statement. Possible violations are:
The EXTEND or OUTPUT phrase was specified but the file would not support write operations.
The I-O phrase was specified but the file would not support the input and output operations permitted.
The INPUT phrase was specified but the file would not support read operations.
I suspect the third, because the relevant bits of my program are :
*----------------------
INPUT-OUTPUT SECTION.
*----------------------
FILE-CONTROL.
* input file 1
SELECT INPUT-1-FILE
ASSIGN TO EXAMPLE
ORGANIZATION IS RELATIVE
ACCESS MODE IS DYNAMIC
RELATIVE KEY IS INPUT-1-ACCESS-KEY
FILE STATUS IS INPUT-1-FS
.
****************
DATA DIVISION.
****************
*--------------
FILE SECTION.
*--------------
* fichier 1
FD INPUT-1-FILE
.
01 INPUT-1-LINE.
05 filler PIC X(300).
*-------------------------
WORKING-STORAGE SECTION.
*-------------------------
77 INPUT-1-ACCESS-KEY PIC 9(3) comp value 1.
01 INPUT-1-FS PIC 99 value 00.
*********************
PROCEDURE DIVISION.
*********************
OPEN INPUT INPUT-1-FILE
DISPLAY INPUT-1-FS
And the DD card in my JCL looks like :
EXAMPLE DD DISP=SHR,DSN=MY.DATASET.NAME
We're using COBOL v5, so I checked IBM's relevant docs, but I can't find any reason that my file is not openable in input mode.
Here are my dataset's characteristics:
And the result of a VSAM listcat:

Answer 1

In your program, ORGANIZATION IS RELATIVE means the input file must be a relative record data set (RRDS) VSAM file, which can be defined using the IDCAMS DEFINE command. What could be happening is that if the file you open is a sequential file, you get a file status of 37. More information about IDCAMS DEFINE is mentioned in IBM Redbook "VSAM Demystified" section 1.5.3, 'Relative record data set'.
You can make the input file a VSAM file with something like the following:
//[YOURID] JOB ,
// MSGCLASS=S,REGION=0M,COND=(9,LT),NOTIFY=&SYSUID,TIME=(1,1)
//*********************************************************************
//* Create a VSAM
//*********************************************************************
//S1IDCAMS EXEC PGM=IDCAMS
//SYSPRINT DD SYSOUT=*
//SYSIN DD *
DELETE ([YOURID].P00695.VSAMRRDS) CLUSTER PURGE
DEFINE CLUSTER (NAME([YOURID].P00695.VSAMRRDS) -
NUMBERED -
RECORDSIZE(80 80) -
BUFFERSPACE(2048) -
SHAREOPTIONS(4 3) -
VOLUMES(SYS162)) -
DATA (NAME([YOURID].P00695.VSAMRRDS.DATA) -
TRACK(1,1) -
CISZ (1024))
/*
//S2REPRO EXEC PGM=IDCAMS
//SYSPRINT DD SYSOUT=*
//VSAM DD DISP=SHR,DSN=[YOURID].P00695.VSAMRRDS
//SYSPRINT DD SYSOUT=*
//SYSIN DD *
REPRO OUTFILE(VSAM) INFILE(INREC)
/*
//INREC DD *
XXXXXXX060ALISONALISONALISONALISONALISONALISONALISONALISONAL ISONALIS
XXXXXXX060ALISONALISONALISONALISONALISONALISONALISONALISONAL ISONALIS
/*

Answer 2

In the IBM docs linked in the question (page 126: FILE-CONTROL paragraph) it is said that the relative access mode is only available for VSAM files :
Problem: the input file is currently not a VSAM file.
Solution: make the input file a VSAM file.
This worked:
DD card in the JCL when creating the input file:
//FOO DD DISP=(NEW,CATLG,DELETE),
// SPACE=(TRK,(1000,1000),RLSE),
// LRECL=300,
// RECORG=RR,
// DSN=MY.DATASET.NAME
The RECORG parameter makes the dataset an RRDS. The lines don't have an easily defined key, so I couldn't use a KSDS and I need the random access, so no ESDS either.
DD card for reading the dataset:
//BAR DD DISP=SHR,DSN=MY.DATASET.NAME
And then in the COBOL program (in FILE-CONTROL):
SELECT INPUT-FILE
ASSIGN TO BAR
ORGANIZATION IS RELATIVE
ACCESS MODE IS DYNAMIC
RELATIVE KEY IS INPUT-FILE-KEY
FILE STATUS IS INPUT-FILE-STATUS
.

Answer 3

If you want to open a sequential dataset for read, you need to open it for Input (done!), have it in the JCL with DISP=SHR (you could try with OLD, but don't need it), and have ACCESS IS SEQUENTIAL.
ACCESS IS RELATIVE as noted by other users is for VSAM only, and your IDCAMS output shows that this is not a VSAM file. If it's just a sequential dataset, you don't need a key either.