Thomas,
"Not all data was written" is a generic error message that does not tell
anything about the actual error. It is issued, when saveproc() returns less
bytes than expected:
eval rc = saveproc( peFD
: %addr(wwData)
: wwLen )
if rc < wwLen
callp SetError(HTTP_RDWERR
: 'errno is currently '
+ %trim(%editc(errno:'L')))
callp SetError(HTTP_RDWERR:'recvdoc: saveproc:'+
' Not all data was written!')
return -1
endif
The actual error must be within the custom saveproc() that you specify at
http_url_post_raw(), parameter "peProc".
I suggest to check the job log for additional information and to debug your
saveproc().
Is it possible that your saveproc() saves the data to an IFS file and that
you do not properly close the file? This way you would run out of file
handles, which is an unrecoverable error for that you need to end the job.
I am just thinking, because you mentioned that after the first error it
"occurs for all the following requests". Running out of file handles might
be one of these errors.
Regards,
Thomas.
Am 03.02.2015 um 11:25 schrieb HIRSCH Thomas:
Hi,
We use httpapi for years with success.
We meet now something very strange: a rpgle program sends successfully
hundred of https requests using "http_url_post_raw" for years
We need to amend this program and now the request is in an internal
procedure (same issue with a subroutine).
The program runs well until a few hundred requests are sent and then
the error "recvdoc: saveproc: Not all data was written§" occurs
When it occurs once then it occurs for all the following requests
SetError() £13: HTTP/1.1 200 OK
recvdoc parms: identity 223
header_load_cookies() entered
recvdoc(): entered
SetError() £0:
<?xml version="1.0"?>
<ncresponse
orderID="E0P4860263354"
PAYID="1333249962"
PAYIDSUB="1"
NCSTATUS="0"
NCERROR="0"
NCERRORPLUS="!"
SetError() £16: errno is currently 0
SetError() £16: recvdoc: saveproc: Not all data was written§
http_close(): entered
Any idea ?
Best regards
Thomas
-----------------------------------------------------------------------
This is the FTPAPI mailing list. To unsubscribe, please go to:
http://www.scottklement.com/mailman/listinfo/ftpapi
-----------------------------------------------------------------------
-----------------------------------------------------------------------
This is the FTPAPI mailing list. To unsubscribe, please go to:
http://www.scottklement.com/mailman/listinfo/ftpapi
-----------------------------------------------------------------------