caBigBox Configuration
In order to test caBIG applications on the test caGrid network, a machine needs to be accessible via the Internet. The following notes document the modifications made to the caGrid Developer VM to achieve this.
Rough Notes
These will naturally be refined.
create a vm dir and unpack the cagrid instance
ssh jpr@meter sudo su - sandefur mkdir vm cd vm tar -xzf tar -xvzf /groups/lab/vmware/vol2/cmbellah/cagrid-0.2.tar.gz cagrid/ exit exit
renamed vm display name to cagrid-sandefur to avoid confusing in vm listing.
was able to load the vm from vmware-server-console as root. but got an error about accessing the user preference file. don't know the impact of this
tasks:
- config machine with better root
- config machine with auto updates
- config machine with gtlib mirror
- config machine with blazerid auth (no)
- config machine with vmware tools
note: should not enable ssh access to machine because of the week password protection of the default vm.
note: when starting the vm as root it runs as the user owner, so it should operate correctly.
bug: (ticket:55) the vm doesn't have networking configured. cmbellah indicated this happened to her and sandefur when they chose "vm was copied" instead of "vm was moved"
looking on-line found a nice description of the source of the problem. basically the linux udev config remembers the original mac address.
source of lost eth0 on vm copy: http://yoopergeek.blogspot.com/2007/07/vmware-loosing-eth0-after-youve-copied.html
the fix for suse is to go into yast->network devices->network card and then you will see two entries with the name "AMD PCnet - Fast 79C971", these are the devices for the virtual network interface from VMware. One entry should list "Not configured" as the IP Address and the other should list "DHCP". The one that lists DHCP is the virtual network card with our original ethernet addess. After the copy of the VM and new ethernet number was assigned to the network device. Delete the card with DHCP configured. Leave the other card unconfigured. Close the network card configuration tool and reboot the vm. this should allow the device to be rediscovered as the only device on the next boot.
can't delete the last remaining network interface in yast. don't know how to force the box to do a device rediscovery. following steps in the post but the only one that applies is to delete /etc/udev/rules.d/70-persistent-net.rules
now start network config. should see just the one device we have. select configure. make sure DHCP is configured for the address specification, and the click next and done. this will start the interface and get it operational.
opps, one more reboot to get this into the "bridged" vm network configuration. with the NAT configuration it doesn't seem the restart of the network interface got eth0 configured with an IP.
bug: the installed language seems to be English (UK)
configured online update through yast but left of the two info-sharing checkmarks.
running vmware-config-tools.pl produces the error about the running kernel and need to build but don't have C or kernel-source installed. select "Base Developmnet" "C/C++ Development" and "Linux Kernel Development" packages. Get ready to absorb 448Mb of data. will be helpful in the future to have this package pre-compiled. granted on non-vm platforms this is not needed.
bug: cagrid vm is missing vmware tools installed. building the modules on a distinct box and moving it to the cagrid release box will avoid needing to install the whole developer suite. though this may not be such a big issue, since this is meant to be a developer box as is. have the size comes from the kernel-source.
vmware-toolbox is running. it didn't install the fast network driver because it wouldn't build.
turn off unneeded services
chkconfig sshd off caGrid:~ # chkconfig cups off caGrid:~ # chkconfig portmap off caGrid:~ # chkconfig postfix off caGrid:~ # service sshd stop Shutting down SSH daemon done caGrid:~ # service cups stop Shutting down cupsd done caGrid:~ # service portmap stop Shutting down RPC portmap daemon done caGrid:~ # service postfix stop Shutting down mail service (Postfix) done caGrid:~ # service ntp stop Shutting down network time protocol daemon (NTPD) done caGrid:~ # chkconfig ntp off
after these services are turned off, i'm only left with mysql and avahi-deamon running. it should be sufficient to expose the box now, since there's nothing really to be seen.
Check open port profile
checking with an nmap of the host
oddly the network interface was down. probably due to the driver rebuild attempt. restarting network didn't seem to fix it so went with reconfigure of device in network services.
nmap of 10.0.0.26 from meter reports that mysql is availble. looks like the fire wall is down.
configure device to external firewall zone and remove ssh from allowed services in that zone. restart firewall.
now nmap is reporting now services available, and taking abit longer due to the firewall dropping (i assume).
nmap results from first and second run:
[root@meter ~]# nmap 10.0.0.26 Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2008-07-01 14:26 CDT Interesting ports on 10.0.0.26: Not shown: 1679 closed ports PORT STATE SERVICE 3306/tcp open mysql MAC Address: 00:0C:29:19:3E:CF (VMware) Nmap finished: 1 IP address (1 host up) scanned in 0.461 seconds [root@meter ~]# nmap 10.0.0.26 Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2008-07-01 14:30 CDT Interesting ports on 10.0.0.26: Not shown: 1679 filtered ports PORT STATE SERVICE 113/tcp closed auth MAC Address: 00:0C:29:19:3E:CF (VMware) Nmap finished: 1 IP address (1 host up) scanned in 22.267 seconds
interestingly the second nmap detects something at port 113 though nothing is running on the host.
bug: vmware-toolbox seems to have died on the cagrid desktop.
change hostname and dns settings in yast
host is cagrid-sandefur.lab.ac.uab.edu and dns is static to official uabhosts.
also changed IP address, the askit update was complete
task
rebooting once more to make sure device comes up properly configured.
updating system
hadn't set the default route
waiting for the vm to update
bug: cagrid instance needs to be preconfigured with updates applied and auto-update turned on.
bug: cagrid instance should have a firewall defined and active
shut down vm and created tarball of current config
