On of the things I almways try to do is to set the EQL LUN’s to Round Robin as Multi Pathing Policy.
- For Existing LUN’s use following PowerCLI script
- For NEW LUN’s
- Install VMware vCLI
- Go to the vCLI command line
- Execute following commandline (you need to enter the password)
esxcli –server SERVERFQDN –username root nmp satp setdefaultpsp –psp=VMW_PSP_RR –satp=VMW_SATP_EQL
–> now each new Equallogic LUN will have Round Robin as default multipathing policy
Some software vendors are licensing there product according the amount of CPU’s there are in you server (Physical or Virtual).
Each vCPU you add in a VMware Virtual Machine is seen as a kind of “physical socket”. This means that if you add 4 vCPU’s in your VM and you need to install software where the software is licensed on CPU SOCKET, you have to pay for 4 CPU’s.
VMware has a advanced setting that makes this 4 CPU SOCKET machine look like a Quaqcore SOCKET, which means only 1 CPU SOCKET.
How to change this setting? Look at VMware KB article 1010184
In my home lab I have a setup with has 1 VC VM and 2 ESXi hosts (Dell PowerEdge T410) which are connected to a Equallogic PS5000 SAN.
The VC and the VC database are both on a different VM.
While I was debugging a powerCLI script for a customer I noticed that my vCenter Service stopped working (and tried to restart automaticaly) without success.
In the Eventviewer I saw following errors:
- In the System Log
The VMware VirtualCenter Server service terminated unexpectedly. It has done this 1 time(s). The following corrective action will be taken in 300000 milliseconds: Restart the service.
- In the Application Log
An unrecoverable problem has occurred, stopping the VMware VirtualCenter service. Error: Error[VdbODBCError] (-1) “ODBC error: (42000) – [Microsoft][SQL Native Client][SQL Server]The transaction log for database ‘VC’ is full. To find out why space in the log cannot be reused, see the log_reuse_wait_desc column in sys.databases” is returned when executing SQL statement “UPDATE VPX_DATASTORE WITH (ROWLOCK) SET NAME = ? , STORAGE_URL = ? , CAPACITY = ? , FREE_SPACE = ? , TYPE = ? , DATACENTER_ID = ? , IORM_ENABLED = ? , IORM_CONGESTION_THRESHOLD = ? WHERE ID = ?”
- In the C:\Users\All Users\VMware\VMware VirtualCenter\Logs I saw erros like “Connection refused”
The answer to all this was lack of space on the drive the vCenter database was located. This because SQL Recovery option on the vCenter database was set to “Full Recovery” and there was NO maintenance plan active. This was the cause of a growing LOG file. After running the maintenance plan the log was emptied and a lot of disk space became available.
At a customer site I have an ESXi server which give me the possibility in the Memory configuration to change the Service Console memory size. But as you know in ESXi there is no more Service Console.
Before I changed something on this server I have put it in maintenance mode. This is a server in Production.
Steps I did without result:
- Restart of the management agents did not fix the issue
- Restart of ESX did not change any thing, setting still available.
- Disconnected the server from vCenter, still Service Console Memory available.
- Reinstalled vpx Agent on the ESXi using VMware procedure (see VMware KB 1003714), still no fix.
This last step fixed the issue
Removed the ESXi from vCenter and added it again. This fixed the issue. Now it looks like a normal ESXi server.
Now we know ESXi is the way to go we need to start to trancition all our ESX servers to ESXi. This is said quicker then done. During a certain period we have build a lot of experiance managing these ESX servers. On of the biggest changes is that there is no more Service Console to login to to watch (using tail / less / more) the log files.
There are now some other possibilities to watch your log files.
Via the DCUI (Consolee of the ESXi). Press F2 (and the password to get to the different options
Using following URL http://<insert ESXi host name or IP here>/host
Using the vSphere Client direct to the ESXi
Using the vSphere client connected to the vCenter and then select the ESXi server you want to export logs from.
Export the logs using PowerCLI using the cmdlet get-log
And the last way is redirect all logs to a Syslog server. Here some examples of Syslog servers you can use. Depending on the needs you can use the FREE version of these syslog servers
Vmware vMA Appliance www.vmware.com/support/developer/vima/vima40/doc/vma_40_guide.pdf#page=26
KIWI syslog server (Windows Service)
SyslogAppliance (Linux Appliance)
www.splunk.com/ (Linux Appliance)
Other suggestions of nice syslog setups are welcome and will be added in here.
VMware is working on a new SRM version in the labs. if you want to know more on SRM goto VMware SRM
I will give you here some product higlights (can change because product is still in devellopment.
* both sites (protected and recovery) are visible in 1 interface. Communication will be encrypted between sites. Certificates can be used.
* Array Managers for both sites can be configured from within one SRM UI. There will be tight integration towards storage.
* SRA interface will be available for each adapter. This will give you a overview of the arrays.
* Array Diagnostics is available to have an overview on storage connections
* Protection groups can be organized in folders
* Protection groups are available for both sites. This to enable failback.
* Protection groups will give of status of individual VM’s.
* Recovery plans are available for both sites to support failback.
* Quick summary for recovery plan and last operation.
* VM startup order can be set in recovery plann
* You can create VM dependencies. Can also start scripts before and after a VM poweron
* IP Custimazation can now be set in VM configuration. You can set een ip config for protected and recovery site.
* Also IPv6 support.
* Possible to do a planned failover. This will assure no data loss dureing migration. This is possible because of tight integration with storage.
* Dashboard will show evolution of the failover on VM level.
* after a failover a Reprotect can be done so a failback can be done without losing data. SRM will tell to the storage to start replication the the other way. A status can be follwoed.
Looking at all these new features, we can hope this version is launched very soon??
Nice work VMware, go on.
here my first blog from VMWorld 2010 San Francisco.
At Sunday before the conference started I had the opportunity to have a look at the Lab room? Duncan Epping has written a nice article about the Lab Rome located in Moscone West. Have a look here http://www.yellow-bricks.com/2010/08/29/vmworld-2010-labs-are-the-place-to-be
Here some facts on the lab:
- A in house written tool called “LabCloud” will be used to manage everything.
- LabCloud uses 12 labmanager environments and some vCenters to provision a lot of vm’s.
- There will be about 30 labs available The lab list
- 480 concurrent sessions can be done.
- about 5000 vm’s each hour will be reprovisioned
- the lab is open for a total of 40 hours
- predictions are that at the end more the 70k vm’s will be provisiond during this week.
- Lab is based on a hybrid cloud, 1 datacenter local and 2 in external datacenters
- Paperless office approach, 2 screens, 1 to do the lab, 1 with the lab manual. And evaluations will be online.
If you are in San francisco, you have to labs to get some handson.
If you are not in SF and you come to VMworld EMEA, you can see the same setup over there (only not that many places)
Hope you will enjoy.
trip to San Francisco
Teaching VMware Fasttrack
Here my first blog. I am gonna try to make some more blogs in the near future. These blog mostly will go about virtualization using VMware products.
To introduce myself look at the about page.
Hope I will you will enjoy my blogging.