Copy MSD directly to and from IFS file

Prior to IBM i 7.2, if you wanted to copy a Main Storage Dump (MSD) to a binary file, you had to follow a multi-step process involving virtual tapes and image catalogs.

At IBM i 7.2, a new set of commands were created that significantly simplifies the process to copy the MSD to a file, ready for transmission.

QMGTOOLS (IBM i Must Gather Tools) has this function built into it, and also can streamline the transmission of the MSD to IBM. Obtaining QMGTOOLS: https://www-01.ibm.com/support/docview.wss?uid=nas8N1011297 Using QMGTOOLS to collect and transfer a MSD to IBM: https://www-01.ibm.com/support/docview.wss?uid=nas8N1020800

For automated handling by IBM, the file naming convention should contain the case number (TSxxxxxxxxx), and a MSD extension ( .MSDxxx).

The IBM i command to copy an MSD entry to IFS directly is:

CPYFRMMSD DUMPID(*SELECT {or dump number}) TOSTMF('/IFS_Path/MSD_Filename.MSD001') REPLACE(*NO {or *YES})

Prerequisites for the CPYFRMMSD command:

• The TOSTMF parameter IFS directory path must exist. If not, you will get CPFA0A9 'Object not found.' The IBM i command to create a directory is: CRTDIR DIR('/IFS_Path/')
• If the default REPLACE(*NO) is specified, then an IFS file with that exact path/name cannot exist. If a file exists already you will get CPF7219 'Stream file already exists and REPLACE parameter value is *NO.'

Examples:

• 1) For case TS000123456, to select the MSD from a list and to copy it to a newly created IFS directory /MSD:

  CRTDIR DIR('/MSD/')

  
  CPYFRMMSD DUMPID(*SELECT) TOSTMF('/MSD/ts000123456.MSD001') REPLACE(*NO)
  
  2) For case TS000123456, if you know the MSD is number 85 in the MSD manager, and want to copy it over the previous ts000123456.MSD001 in the IFS directory /MSD:
  CPYFRMMSD DUMPID(85) TOSTMF('/MSD/ts000123456.MSD001') REPLACE(*YES)

  note : The directory /MSD already existed in example 2, so didn't need to be created

 

Thinking of zipping the MSD file or saving the MSD file to a SAVF?

IBM i MSDs are already compressed. Zipping them or saving them to an IBM i SAVF creates extra work for you and for IBM Support, delaying the analysis of your data. Additionally, zipped or SAVF MSDs will not perform time-saving automation, as discussed in the next section. This further delays analysis.

Procedures for transmitting an MSD to IBM for review

For fastest processing, it is recommended that you put the case number at the beginning of the filename. For example, if your case number is TS000123456, we recommend the TOSTMF parameter be:

TOSTMF('/ExistingIFSdir/ts000123456.MSD001')

Doing this properly will automatically connect this MSD to the case and its data repository at IBM. Additionally, the automated process will move the MSD to the appropriate release system, ensuring the fastest possible review time.

Now, you are ready to start an FTP session to testcase or ECuRep and send in those files:

Choose the upload server that is closest to the location from where you are going to upload the MSD:

Americas: testcase.boulder.ibm.com (IP 170.225.15.31)
Europe, Middle East and Africa: ftp.ecurep.ibm.com (IP 192.109.81.7)
Asia Pacific: ftp.ap.ecurep.ibm.com (IP 210.143.141.69)

Initiate an FTP connection, and login with 'anonymous' and your e-mail address as the password.

NOTE: At some point in time, IBM will remove anonymous login. Refer to this URL for information to obtain an IBM ID and password.

http://www-01.ibm.com/support/docview.wss?uid=nas8N1022576

Issue the following commands:

namefmt 1
bin
lcd /ExistingIFSdir
cd /toibm/os400
mput ts000123456.MSD001

Remember, /ExistingIFSdir and ts000123456 are just examples, substitute them with appropriate directories and case numbers, or the commands won't work, and your MSD analysis may be delayed.

Is the dump too large to FTP all at once? Split it up.

You should try to determine how much data can be reliably transferred without interruption. For example, if you able to get to approximately 2000 MB before the transfer fails, pick a split size of 1500 MB. Let's call this the split size, SSSS, which is a whole number of Megabytes (no decimal places).

The 'split' command has the limitation of only being able to split into a maximum of 676 pieces ('aa' through 'zz'). Needing to split into that many pieces is a good indication that it would likely be faster to send a copy of the MSD on physical tape instead (refer to Rochester Support Center knowledgebase document Copy MSSD and LIC Logs to Tape at SST - RISC: http://www-01.ibm.com/support/docview.wss?uid=nas8N1018368).

To determine how many split pieces you will create, divide the original MSD IFS file size (in bytes) by 1048576 (1 MB), divide that by the split size (SSSS) , and then add 1. If this number is greater than 676.0, you cannot split the MSD file, and the MSD needs to be sent on physical tape media (refer to Rochester Support Center knowledgebase document Copy MSSD and LIC Logs to Tape at SST - RISC: http://www-01.ibm.com/support/docview.wss?uid=nas8N1018368).

You can perform the split command through PASE ('call qp2term'). QSHELL is 32 bit-based and has file size limitations:

SBMJOB CMD(CALL PGM(QP2SHELL) PARM('/QOpenSys/usr/bin/-sh' '-c' 'split -b SSSSm /ExistingIFSdir/ts000123456.MSD001 /ExistingIFSdir/ts000123456.MSD001_')) JOB(SPLIT_MSD) JOBQ(QSYSNOMAX)

Once this completes, you need to record a few pieces of information for IBM Support to ensure proper joining of the files.
i) Record the first split file name, and size in bytes.
ii) Record the last split file name, and size in bytes.

This will tell IBM Support the exact split size (size of first file), how many files to expect (name of last file), and the size of the last split piece which is likely less than the split size.

Once the transfer is complete, you should relay the original file size, the first split file name and its size in bytes, the last split file name and its size in bytes.
 

---

If there is a need to restore this MSD back into the MSD Manager (on this same system or another), you would use the new command:

CPYTOMSD FROMSTMF('/IFS_Path/MSD_Filename') DESC('Descriptive MSD text in 32 chars')

Prerequisites for the CPYTOMSD command:

Note: The CPYTOMSD command is only used to load an IFS MSD image file to the main storage dump manager in service tools. This option is typically used to review dumps on another system and is not part of the process to send a dump to IBM.

• The FROMSTMF parameter IFS full path must exist. If not, you will get CPFA0A9 'Object not found.'
• The FROMSTMF parameter IFS file must be an MSD file created using the CPYFRMMSD command. You can't use it to restore a virtual tape image of an MSD. If CPYTOMSD is attempted with a file that is not valid, you will get CPF7221 ‘Error occurred copying main storage dump.’
• The DESC parameter must be 32 characters or less, or not specified. If the DESC parameter is more than 32 chars, you will get CPD0074 'Value 'xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx' for DESC exceeds 32 characters.