Short lifespan on SD cards booting ESXi 6.0 U2.
Client was seeing failure rates on SD cards much higher than average. They quickly ruled out a manufacturing flaw as both their HPE ProLiant and Dell PowerEdge servers were seeing similar failure rates.
They did not properly configure a ScratchConfig location and by default the boot disk (SD Card) was being used. This is find when using hard drives or SSD’s but SD cards aren’t designed to take the constant disk traffic. The ScratchConfig should have been redirected to their SAN storage. To do this
- Launch the vSphere Client
- Create a small data store just for your hosts to use. You can use any data store but I always use a dedicated one just for management. For the sake of brevity, I’m going to assume you know how to create a data store and how to find it’s location path.
- In that data store, create a folder with the ESX server name. I typically use .locker-servername.domainname.com
- Select an ESXi host
- Place the host into maintenance mode.
- Click on the Configuration Tab
- Click on Advanced Settings
- Go to Scratch Config
- Locate ScratchConfig.ConfiguredScratchLocation and enter a path to your data store. It should read something like this: /vmfs/volumes/98a66830-add1f5c5-f487-44a847777a22/.locker-servername.domainname.com
Again, it’s not necessary to have your own data store for this but I really like having it separated out.