UPDATED: ExampleVM scripts and documentation available

[ChrisFloyd]

Tom,

Yes, that will work. 

[tdeneau]

When running 2 tiles using the Example VMs, I am trying to understand what to do with the clients.
Can I not drive the two tiles from a single client with that client having the full for set of IP addresses
for all the VMS in the 2 tiles?   I changed Control.config to have NUM_TILES=2
but when I start runspecvirt.sh, it is looking for a host called client2.

-- Tom

The script documentation states that the goal of the scripts is to set up a single tile. We're glad to hear you were successful. I considered writing a script to automate this, but since you're in a hurry, we wrote up the steps instead.

To clone a tile, you need to do the following.

• Change the hostname (/etc/sysconfig/network) and IP addresses (/etc/sysconfig/network-scripts/ifcfg-eth0 and ifcfg-eth1).
• Delete /etc/udev/rules.d/70-persistent-net.rules if it exists.
• Edit the hosts file (/etc/hosts) to reflect the new hostname and IP addresses.
• Reboot the VM. Log in and make sure you can ping other VMs.

Extra for appserver:

• Edit the hosts file (/etc/hosts) to make sure that the alias specdb points to the correct dbserver (up to four appservers for each dbserver). That is, the specdb alias for appserver1-4 points to dbserver1-int. The specdb alias for appserver5-8 points to dbserver2-int.

Extra for infraserver:

• Edit /etc/exports to change the tile number for webserver.

Extra for webserver:

• Edit /etc/fstab to change the tile number for infraserver. (You may want to use a VM internal network for this mount if available.)
• Make sure that the support/downloads directory is mounted from the infraserver VM via NFS.
• Go to the /opt/SPECweb2005/wafgen directory and edit the support wafgen files to set the TILEINDEX value for the tile you are creating:
  
• Edit /opt/SPECweb2005/wafgen/unix/support_downloads_props.rc and set the parameter TILEINDEX equal to the tile number minus 1 (for example, set TILEINDEX=1 for tile 2).
• Edit /opt/SPECweb2005/wafgen/unix/support_image_props.rc and set the parameter TILEINDEX equal to the tile number minus 1 (for example, set TILEINDEX=1 for tile 2)
• Build the image files: ./Wafgen unix/support_image_props.rc
• Build the support files: ./Wafgen unix/support_downloads_props.rc

[tdeneau]

I think I have followed all the steps outlined below for tile 2 webserver/infraserver, but when I try to access     

http://webserver2/support thru a browser I always get
    "ScriptError: BeSim returned with an error number: 1"
My webserver/infraserver works fine on tile 1.
Any suggestions?

-- Tom

Extra for webserver:

• Edit /etc/fstab to change the tile number for infraserver. (You may want to use a VM internal network for this mount if available.)
• Make sure that the support/downloads directory is mounted from the infraserver VM via NFS.
• Go to the /opt/SPECweb2005/wafgen directory and edit the support wafgen files to set the TILEINDEX value for the tile you are creating:
  

• Edit /opt/SPECweb2005/wafgen/unix/support_downloads_props.rc and set the parameter TILEINDEX equal to the tile number minus 1 (for example, set TILEINDEX=1 for tile 2).
• Edit /opt/SPECweb2005/wafgen/unix/support_image_props.rc and set the parameter TILEINDEX equal to the tile number minus 1 (for example, set TILEINDEX=1 for tile 2)

• Build the image files: ./Wafgen unix/support_image_props.rc
• Build the support files: ./Wafgen unix/support_downloads_props.rc

[/list][/list]

[ChrisFloyd]

Tom,

Technically, it is possible to run multiple clients on the same physical machine (i.e., OS), however, this can quickly become a complex and frustrating environment to maintain for more than a few tiles.   This is because the ports used for WORKLOAD_CLIENTS[Tile x][ Workload y], and the CLIENT_APP[Tile x][workload y] must all be unique. I suspect this may be the problem you are encountering. Most folks find using virtual clients (which can be cloned, replicated, and then edited only for /etc/hostname numerical IPs) a much simpler implementation.

[tdeneau]

Thanks, Chris.
Yes, since posting this, I have switched to running each client in its own VM now.

My main problem seems to be the webserver on tile 2.
Digging down, I see that there is no /tmp/besim_support.globals on infraserver2 (unlike on infraserver1).

[tdeneau]

Meant to also add, using tcpdump, the error message I see coming back from BeSim is ""access to globals failed"

In a related question, even on infraserver1 I only see besim_support.globals, not besim_ecommerce.globals or besim_banking.globals.

-- Tom

Thanks, Chris.
Yes, since posting this, I have switched to running each client in its own VM now.

My main problem seems to be the webserver on tile 2.
Digging down, I see that there is no /tmp/besim_support.globals on infraserver2 (unlike on infraserver1).

[tdeneau]

I noticed that if I copy /tmp/besim_support.globals from infraserver1 to infraserver2 and chown to apache:apache, the webserver2/support page does come up without an error.  Still have to see if the whole workload passes.

-- Tom

My main problem seems to be the webserver on tile 2.
Digging down, I see that there is no /tmp/besim_support.globals on infraserver2 (unlike on infraserver1).

[johnnyli]

The latest v1.03 example has the note it does not support Centos7.x and RHEL7.x as below.
NOTE: The scripts and steps described in this guide work with CentOS 6.x and RHEL 6.x. They do not work with the systemd architecture-based CentOS 7.x or RHEL 7.x.

But in the script there are many systemctl command not supported by 6.x which is enabled in Centos and RHEL from 7.X only.  As a result, i can not try the example in Centos 6.x and RHEL 6.x

What I misunderstand here?
Can I use this example for SPECvirt_sc2013-1_1p1.iso?

[johnnyli]

Download v1.04 and failed to uncompress  the tgz file:(


[root@mailserver Downloads]# tar xzf optdirs-sc2013-v1.04.tgz
tar: This does not look like a tar archive
tar: Skipping to next header
tar: Exiting with failure status due to previous errors
[root@mailserver Downloads]# tar xzf helper-sc2013-v1.04.tgz
tar: This does not look like a tar archive
tar: Skipping to next header
tar: Exiting with failure status due to previous errors
[root@mailserver Downloads]# ls -l
total 8
-rw-r--r-- 1 root root 1820 May 31 00:47 helper-sc2013-v1.04.tgz
-rw-r--r-- 1 root root 1663 May 31 00:47 optdirs-sc2013-v1.04.tgz
[root@mailserver Downloads]# md5sum optdirs-sc2013-v1.04.tgz
d0f5e7817c6e611f8923753ee140f2d4  optdirs-sc2013-v1.04.tgz
[root@mailserver Downloads]# md5sum helper-sc2013-v1.04.tgz
ff0a24bf0fe9d71cfcba84374d50bf59  helper-sc2013-v1.04.tgz
[root@mailserver Downloads]#

[lroderic]

These are good - sorry for the delay. Please retry.

[johnnyli]

yes, it works now. I'm playing with it. Will come back to you in new topic thread for any issue, thanks in advance

[Leslierichardson]

Include a README file that contains. Allow issue tracker for others. Write an API documentation. Document your code. Apply coding conventions, such as file organization, comments, naming conventions, programming practices, etc. Include information for contributors.