I stopped my Amazon EC2 instance with Teradata developer edition (4AMP) to prevent from being charged (except for Volume). Since I turned it on, Teradata database server hasn't come up or at least not letting me connect. I have tried restarting the Teradata services (tpa, pdestate check) but I keep getting the error "*** Warning: RDBMS CRASHED OR SESSIONS RESET. RECOVERY IN PROGRESS." Have stopped all Teradata services, restarted the EC2 instance and started the Teradata services again but no luck. Has Teradata database server got corrupted? Is there a way we can get it working or any repair option available?
Teradata devleopers edition uses SSD local storage. Local storeage is lost on stopping it.
Local storage is temporary or ephemeral and is ideal for testing and development where you do not need to keepdata. Local storage uses SSD or HDD media.
You can find more information on stoping and starting your instance from AWS documentation http://docs.aws.amazon.com/AWSEC2/latest/UserGuide
Ok, that kind of makes sense. When I run verify_pdisks it's resulting in disk errors and reports that it cannot find directory/file. Does that mean my Teradata install is gone? What is concerning is I have now effectively lost all data which I am not worried about but can't seem to get the Teradata server working again to do any work. Is there any recommended way to get this fixed or should I just trash this and commission a new instance and install Teradata afresh?
Output from verify_pdisks
Error: Cannot open /dev/pdisk/dsk1.
Error: Cannot open /dev/pdisk/dsk3.
Error: Cannot open /dev/pdisk/dsk0.
Error: Cannot open /dev/pdisk/dsk2.
Once you stop a Teradata Database Instance with Local Storage, the data is lost and database won't come up as all the underlying disk storage has been taken away. You need to terminate the broken instance and start an new instance and configure Teradata database again.
Please refer the below link for more info,
All the above is correct - you do lose all the database data (including metadata).
However, we did find that you can keep the same instance, bring it up, stop the database (because it thinks it's still running but it really isn't and then run tdc-init -force. This will force the initialisation of the database on that instance.
At least with this approach you don't need to go right back and create a new instance and do all the networking configuration.
Hope this helps.