Installation Guide

See the System Requirements to ensure you meet the requirements before installing. These instructions have been written with RedHat in mind so your Linux environment might force changes (e.g. different cdrom mount point)

Standard Install

Insert the simulator cdrom and mount it onto your Linux host.
```
  [root@localhost /root]# mount /mnt/cdrom
  [root@localhost /root]# cd /mnt/cdrom
  [root@localhost cdrom]#
```
If you downloaded the .tgz format you’ll need to unpack it into a directory.
[root@localhost /root]# tar xvfz 641simulator-tarfile.tgz [root@localhost /root]# cd simulator [root@localhost simulator]

Start the setup.sh script

  [root@localhost cdrom]# ./setup.sh
  Script version X (Jul/2003)

Where to install to? [/sim]:This is asking where to install a new simulator. If you press enter you’ll get the default of /sim, otherwise choose a new directory. Do not install into the / root directory! If the directory already exists and doesn’t appear to contain a simulator then you will be asked:
```
<directory> already exists. This may overwrite files.
Are you sure? [no]:
```
Only say yes if you are sure that the directory that you are installing into doesn’t contain anything useful.
When installing if you choose a directory that already has a simulator installed it will ask you if you want to upgrade it. See Upgrading a Simulator.
Would you like to install as a cluster? [no]:

If you specifically want to test some cluster features then say yes. Otherwise (and normally) you want to choose the default answer. See below for additional notes on clustering.
Install full HTML/PDF documentation installed? [yes]:

In a non-clustered install you will be asked if you want the full ONTAP documentation installed. If you say no you will still get the “man” pages and FilerView. Adding full documentation will require an additional ~130MB of disk space.
Continue with the installation [no]:

If you’re happy with the choices you’ve made enter yes. The installation will start. E.g. I chose the default /sim. Here is the output.
```
  Creating /sim
  Unpacking sim.tgz to /sim
  Configuring the simulators mac address to be [be:ef:74:b0:5:e1]
```

At this point the simulator is installed and you are now walked through the simulator configuration steps.

Configuration of the simulator

You can manually run the configuration script afterwards by changing into your simulator folder and entering ./setup.sh.

Regenerate MAC address? [no]

This is only asked if you are reconfiguring. It’s rare that you’ll want to change the MAC address for your simulator, but if you do then enter yes here.
Use DHCP on first boot? [yes]:

The simulator can use DHCP to get its initial IP address. It never uses DHCP again so say no if you are on a public network. If you have a private network (or are using something like VMware in NAT or Host-Only mode) then you can safely say yes.
Ask for floppy boot? [no]:

If you want to run maintenance mode or initialize all the disks then say yes otherwise for normal use say no.
Which network interface should the simulator use? [default]:

If you have more than 1 network interface (e.g. 2 interfaces – eth0 and eth1) then you can tell the simulator what network interface you want it to start against. This is especially useful if you have 2 interfaces on the same subnet – choose the interface that isn’t the default route (e.g. eth1).
If you aren’t sure leave as default and the simulator will choose an interface (usually eth0).
How much memory would you like the simulator to use? [128]:

The simulator uses memory for buffering, maintaining connection information etc. The larger the number you put here the better the simulator will run but you don’t want to starve the host machine of memory. 128MB should be enough for basic SnapMirror – if you don’t need SnapMirror then you can probably reduce the memory and if you are running lots of software (e.g. MultiStore, SnapMirror, Clustering etc) then you may wish to increase the amount of memory the simulator is given. There is an additional overhead of around 32MB per simulator that isn’t taken into account by this question – see top or vmstat to get an idea of the memory usage when it’s running.
For a cluster you should give each node 2x the amount you think each nodes needs. For example if you want to give 80MB to a node really give 160MB so that on a failure the surviving node can function correctly. This means that while both nodes are running you’ll use up a lot of physical memory so make sure your system has enough spare to cope.
Create a new log for each session? [no]:

If you say yes you’ll be given a new log file each time with a semi-random filename based on the number of seconds since 1970. Otherwise you’ll be opting to have just one log file and will be asked another question:

Overwrite the single log each time? [yes]:

Say no here if you want a log file that holds everything you’ve ever done in the simulator. Although it shouldn’t grow too large you might occasionally want to delete/rename the file so that a fresh one is created.
Your simulator has 2 disks. How many more would you like to add? [0]:

By default 1 or 2 disks are installed in your simulator for the root volume. You will probably want to add more disks so you can create your own volumes. When you select more than 0 you’ll be asked:
```
  The following disk types are available in MB:
          Real (Usable)
    a -   43   ( 14)
    b -   62   ( 30)
    c -   78   ( 45)
    d -  129   ( 90)
    e -  535   (450)
    f - 1024   (900)
  What disk size would you like to use? [a]:
```
The simulator supports different disk sizes. The table displayed shows how much space will be available in the simulator (Usable column), how much space it takes up in Linux (Real column) and what letter to press. The disk type used for vol0 was type “d” though remember you can mix and match disk sizes in the same volume/raid group. Bigger disks are always more space efficient.
After choosing the disk size you’ll be asked:
```
Disk adapter to put disks on? [0]:
```
If you are using SyncMirror then you may want to choose adapter 1 for the second plex disks. For all other cases choose adapter 0. See the SyncMirror configuration requirements.
You have X pulled disk(s). Delete them to recover space? [no]:

This question will only appear if you’ve followed these instructions to recover disk space.
The configuration is now complete. A final test will be performed by the script to check for the correct versions of the gcc libraries that the simulator requires. It does this by starting the simulator and observing the error condition that is returned.
```
  Starting simulator to check for Linux segfault issue…
  ignore the next few lines between – marks
  -----------------------------------------
  ./setup.sh: line 13:  1298 Aborted                 ./maytag.L
  -----------------------------------------
  segfault issue not detected.
  Complete. From the simulator directory run runsim.sh to start the simulator.
```
Any error messages output during this stage can be safely ignored. The simulator is now ready for use.
When performing a clustered install you’ll see a small amount of additional information as the script configures the cluster.
Start the simulator. E.g. if you installed it into the /sim directory then do the following:
```
  [root@localhost cdrom]# cd /sim
  [root@localhost /sim]# ./runsim.sh
  

  NetApp Release 6.4R1: Thu Mar 13 23:56:30 PST 2003
  Copyright © 1992-2003 Network Appliance, Inc.
  Starting boot on Wed Apr  9 10:50:25 GMT 2003
  etc…
```
You will see a few messages about uninitialized disks being treated as spares and some ftp option errors, this is normal.
Finally you’ll be shown the standard first-time setup including being asked for host names etc.

If you installed a cluster then you’ll need to run each node of the cluster and turn on the cluster license (see License Keys). See the normal filer documentation for details on configuring and running a cluster.
If you want to change any of the answers to the questions above like the amount of memory for the simulator, run the setup.sh script from the simulator directory. E.g.
```
  [root@localhost root]# cd /sim
  [root@localhost /sim]# ./setup.sh
  Please ensure the simulator is not running.
```

Clustered Install Notes

A clustered install is similar to a non-clustered installation as detailed in the main instructions with the additional considerations.

You need to decide if you will use each node in an active-active configuration, or use “node 1” (the first filer) as your primary test/work area, and keep “node 2” to prove clustering as/when you need it. I generally recommend the later as you will be able to minimise the disk and memory requirements needed, plus you can keep the second node powered off most of the time and not lose access to any of the data that is contained in it.
Each simulator has its own disks. If you follow my recommendation above then you’ll want all your disks allocated to “node 1” and will keep the default minimum number of disks for node 2.
If you need the gcc library fix then you won’t be able to use clustering. This is because the fix includes chrooting into the directory that contains the simulator and has the effect of breaking the symbolic links that the simulator uses to find its partners NVRAM, disks etc. Currently the only solution would be to change to a Linux that doesn’t need the fix unless you’re into hacking together your own chrooted environment that contains the gcc environment from RedHat 7.1.
This is one good reason to VMware for the hosting of the Linux environment that the simulator uses – you can run one version of Linux (or Windows) outside the simulator environment, and a specific version of Linux inside the VMware session.
Give each simulator more memory (2x is a good figure). The simulator will need more memory to run all the services that the failed node had.
You cannot use SyncMirror in a clustered environment. This is due to the emulation a type of cluster interconnect card that SyncMirror doesn’t support.
Before halting one of the nodes remember to cf disable otherwise the remaining node will take over. You may want to do this anyway.
To simulate a dirty shutdown of a node you can press [ctrl-N] and the node will instantly stop.