A lot of people host Linux systems on z/OS mainframes. Standard mainframe backup, recovery and management products like FDRABR, DFDSS or DFHSM cannot be used to manage the data held within these Linux systems, but FDR/Upstream can be installed onto a z/OS server and then used to manage and backup the internal data. This allows you to automate backup and archiving of your Linux data using z/OS batch jobs while utilising your existing mainframe disk and tape drives for storage.
It is quite straightforward to install the Upstream mainframe server. You download all the files from tape to disk, then exec a clist called ustalloc. This takes you through a set of ISPF panels. Upstream needs a set of control files to operate. You take option 3 from the panels to generate JCL to allocate these control files. You need to APF authorize your upstream control library before you can run these jobs.
You need to set up a VTAM applid for Upstream and get it started at IPL time. If you cannot do this, you need to find a VTAM expert to help you, as your VTAM libraries will be site specific.
The Upstream started task will be in a library called *.icl32(ustproc). The * will be the naming convention that you used when you downloaded the files. Update the proc to fit your own naming conventions, then copy it out to an appropriate JES proclib dataset.
You then need to set up the config using option 4 from the ISPF dialog invoked by the ustalloc clist. If a config file already exists, the clist will pre fill the file name in the input dataset name, and will prompt you for a config member name.
The FIRST time you allocate the config file, blank out any pre filled input file name, and just include the output file name, and member. Do not allocate an empty member in the PDS library yourself, as this prevents Upstream from saving the config.
Installation is a simple as copying all the Upstream files to a SYS:/UPSTEAM/ directory, defining a search path to that directory, then loading the NLMs
Upstream uses the following NLMs
The key to getting the Upstream Netware client working, is to get the Netware login profile, and the Backup profile right.
Upstream logs into a Netware server using a 'Netware Profile'. You need to set up a Novell profile for each server, and you do this with the USCFG.exe utility. When you execute it, you see the screen
In this case, the Server Comms is using a DNS entry for the mainframe server, rather than using a hard coded TCP/IP address.
Click on the 'Advanced' button, and you see
This is where you define your Netware profile. As you need a Netware profile for each server, it is a useful convention to use the server name for the profile name. Type in your profile name in the Novell Profiles window, then select 'ADD' and the following screen appears
You need to supply the server name, and a userid and password. The userid must be in 'Typefull Netware' format as shown above. The userid must have sufficient Netware rights to be able to carry out migrate and recall tasks.
For HSM archiving to work, you need a file called nwrecall.dat on the server, and when it is installed and initialised, it will probably have incorrect paths and parameters in it. The data it contains is -
Make sure that the PARAMETER line at the top points to the correct PATH to your Netware server, it may have defaulted to the drive mapping on your PC. Make sure that the BACKUPPROFILE matches the profile you use to run an archive from the mainframe, and that the NOVELLPROFILE line points to a profile which you defined using uscfg.exe as above. As far as I know, the only way to change the nwrecall.dat file is to edit it directly.
Upstream migration works with Netware 6 and 6.5 clusters. It will not work with Netware 5 clusters (but it does work with Netware 5 stand alone servers).
Each NCV has its own virtual server name and IP address. This is explained in the Netware section. You need to setup a Netware Profile for each volume, and each profile must have a TSA name, which is of the format virtualservername.netware clustered directory service.
Each NCV needs its own copy of us.nlm and usrecall.nlm, and these are loaded on whichever server is hosting the volume at the time. This means that each NLM must have a unique name, us1, us2, us3 etc, or nwrecall1, nwrecall2, etc. You need to include this in the server takeover scripts, so when a disk failsover to a different server, the correct NLMs are moved with it.
For example, consider a 2-node cluster with 2 cluster volumes. The nodes are called NCL1 and NCL2, and the volumes are called VCL1 and VCL2. The cluster is in an NDS tree called CLNDS, so the the cluster volumes are associated with virtual servers called CLNDS-VCL1-SERVER and CLNDS-VCL2-SERVER
NCL1 and NCL2 will both run us.nlm and usnds.nlm, if you want to backup your SYS: volume and NDS using Upstream.
VCL1 has us1.nlm and usrecall1.nlm defined. It is currently hosted by NCL1, so these NLMs are running on the NCL1 server.
VCL2 has us2.nlm and usrecall2.nlm defined. It is currently hosted by NCL2, so these NLMs are running on the NCL2 server.
If NCL1 fails disk VCL1 will failover to NCL2, and the takeover scripts will start us1.nlm and nwrecall1.nlm on NCL2 server.
NCL2 will then be running us.nlm, us1.nlm, us2.nlm, usnds.nlm, usrecall1.nlm and usrecall2.nlm.
You need to install a complete copy of Upstream on the cluster disk, in an Upstream directory. This copy will hold the USx.NLM.The USRECALLx NLM code is held on a directory on the cluster disk, called RECALL. You need to create a NWRECALL.DAT file in the Upstream directory, and it must contain the Netware profile for logging into the cluster disk. Each copy of NWRECALL must have its own unique TCP/IP port number.
You can add a backup profile from either the PC or the mainframe component of Upstream. The PC GUI is probably a bit easier to use.
To use your PC GUI, start up us.exe, take 'management' then 'profile configuration' and you will see a screen like this. If the profile list is blank, then put a '*' in the Profiles field, then click on 'profiles' button to list profiles
Type in a new profile name and hit the 'add' button, or select a profile that you want to change, and hit the 'update' button, and you will see the following screen.
Although this is a 'backup profile' it is actually to be used for migration only, as the box at the bottom left is checked. It is also migrating files to mainframe disk.
To change backup profiles from the mainframe, go into the Upstream panels, and take option 5, configuration profiles, and hit enter to see the profile list in config00 as below
Put an 'e' against the NOVMT profile, and you can edit it. ITS IMPORTANT TO KNOW THAT THE DATA DOES NOT ALL FIT ON ONE PAGE, the MIGRATION detail is at the BOTTOM of the page, and the default is NO MIGRATION. Page down with PF8 and see
Make sure PCMIGRATE is set to YES. Don't get hung up about PC migrate, they really mean client migrate, which could be a PC or a file server.
If you change a Backup profile, you need to refresh the mainframe started task using the ISPF command