build a new vmware cluster for a customer:
2*ESXi 4.1@x3650M3, 1*vSphere,Essential Plus@x3650M3, 1*DS3500,2*SAN24B FC Switches. 1*TS3100 LTO5 FC Library
ignore the hp garbage …
upgraded to kernel 3.0.0-1-amd64 on debian wheezy; to get vmware workstation 7.1.4 back running, just install that patch :
a ibm ds4700 with four exp 810 units lost it`s preferred path setting for one controller /some logical drives. consequences : there were some machines with non-redundant paths connected to the san – these machines lost their mounted luns.(datastore for a mail machine, datastore for a fileserver with 14TB…) first i changed on the fibre channel switch fabric the zoneing to the healthy controller and remounted the drives to get the server systems back on production. after some research i seems to be clear that the solution is just only to map the logical drives back to the preferred controller/path … *sigh you can change the controller settings for the logical drives under production scenario. learned: do *all* fc/san connections redundant – don`t save money on your fibrechannel san design … 🙂
message on the storage manager / recovery guru : Logical Drive Not On Preferred Path
today i upgraded my workstation to debian testing aka wheezy with kernel 2.6.38-2-amd64. some problems with the mysql-server-5.1 and with vmware workstation 7.1.4 but overall ok. ( vmware patch : http://www.linuxinsight.com/vmware-workstation-7.1.3-runs-great-on-linux-kernel-2.6.37.html )
just updated some esxi hosts, merged a snapshot with a flatfile vm, switched 17 servers to different esxi hosts – vmware is a good product. now we can do even online backups with esxi 4.1 – ghettoVCB.sh http://communities.vmware.com/
docs/DOC-8760
one important thing left so say: never touch or move vms with snapshots ^^
copied vm centos 5 machines with fixed ip settings to a different esxi server. put the machines in a vm net with an active dhcp server. the centos machines switched at boot-time to dhcp and automatically disabled the fixed ip settings. centos creates a new interface configuration file with dhcp settings. the old config file is backed. why ? the mac adress has changed ! (vmware moving). to prevent this you should probably configure the mac adresses from the old esxi system or adjust the network cfg files on centos.
i`m planning a new virtualization structure for a customer. this time based on three ibm x3650 m3 machines; one machine as iSCSI target, the other two machines with vmware ESXi 4.1, for sure it will be real fun to configure one x3650 m3 as a iSCSI target – hoping debian stable will install clean without patching the kernel for raid modules or something …. 🙂
didn`t turned on VT at my new quadcore workstation and wondered weeks about the slowness of the vmware machines …. *fail
finished a classical virtualization project
used : 4* hp proliant dl380g7, 1* hp storage works p2000 g3 fc + 2 extension, vmware esx 4 vsphere essential plus, hp procurve 2910 al 48g, hp usv r8000
frontside
backside
