[MITgcm-support] failure to find file

[Holly Dail]

Hello all -

My runs on pleiades are failing frequently due to a reported failure to find one of my boundary condition files (I'm using prescribed open boundary conditions), but the files are present and accounted for before the run and after the run failure, and I presume during the run too!  Runs that are short are frequently okay, but one that takes several hours has failed four times in a row.  Basically almost every read is just fine, and then every once in a while the read fails and my run aborts.    This happens even if the experiment has its own copy of the relevant files.

It looks like globalFile is being set to .FALSE. in the following code block from mdsio_read_section.F after the 'INQUIRE' fails to find the file.  Can anyone help with some advice as to making the INQUIRE call more robust ... I suppose I can write a loop and ask several times to make sure 'no' is the right answer, but it doesn't seem like this should be happening in the first place.  My MIT GCM version is checkpoint 62m + some additions (checked out on 11/4/2010).

Thanks for any advice!
Holly

C Check first for global file with simple name (ie. fName)
      dataFName = fName
      INQUIRE( file=dataFName, exist=exst )
      IF (exst) THEN
       IF ( debugLevel .GE. debLevA ) THEN
        WRITE(msgBuf,'(A,A)')
     &   ' MDS_READ_SEC_XZ: opening global file: ',dataFName(1:IL)
        CALL PRINT_MESSAGE( msgBuf, standardMessageUnit,
     &                      SQUEEZE_RIGHT, myThid )
       ENDIF
       globalFile = .TRUE.
      ENDIF