esxi

I’m a bit late to the blogosphere with with one, but we’ve had a couple of customers hit the issue described below, and it seems to be mostly across HPE and Dell hosts running ESXi 6.x. I’ll get to the good stuff first. If you’re hitting an issue on ESXi 6.x where the RamDisk is filling up and ScratchConfig.CurrentScratchLocation is reverting to /scratch, even with a location configured, it’s likely due to a known issue with an Emulex Driver.

I’ve just been working through automating the powering on and off of my lab server, which is a single physical server running ESXi 6.5, and nested virtual machines for my lab. I wanted to use the ESXi Autostart feature to start my domain controller and jump box at power on. But I had an issue where one of the two VMs I want to power on would simply not start.

Quick post. The other day I saw an alert on an ESXi 6.0 host as follows: Agent can’t send heartbeats: No buffer space available The host was responsive and connected to vCenter Server, and the VMs were running OK on the host. I checked the vmkernel / hostd / vpxa logs, and the only mention of this was a one liner in the vmkernel log which repeated the error message above.

Introduction Nimble Storage arrays are one of my favourite SAN arrays to work with at the moment. However, until Nimble OS3 which was made GA on the 31st of August, 2016, there was no support for the XCOPY VAAI primitive. This meant that when performing actions such as a storage vMotion that moved data between datastores on the same array, or cloning virtual machines, the ESXi host performing the action was required to read the data from the array up through the ESXi host, and then write it back to the array, creating additional load on the ESXi host and traffic on the storage adapters and network.

In the busy world we live in, a lot can be said for the “time flies when you’re having fun” saying. And with so much happening in the world of the SDDC, it’s easy to forget about some of the core components that, for the most part, sit there reasonably quietly just ‘doing their thing’. This is just a quick reminder post for those still running vSphere 5.0 or 5.1, that end of general support ends for those products this week, on the 24th of August to be exact.

Introduction As of vSphere 6.0U1, VMware allow an embedded vCenter server deployment to be reconfigured to an external deployment, which demotes the Platform Services Controller (PSC) components of the embedded node and points the VC server to an external PSC node which resides in the same Single Sign On (SSO) domain as the source embedded node. This is done by using the utility cmsso-util Before we get too much further, there are two main uses for cmsso-util: