Our monthly webcasts always generate excellent questions from the audience. And when Francis Daly (one of our Staff Technical Training Specialists who trains VMware’s technical staff on all things storage) delivered Virtual SAN Troubleshooting a few weeks ago the questions were outstanding.
We pulled a few of the questions below, and you can read the entire list and listen to the recording on your schedule.
Q: I have a test vSAN environment with 3 ESXi hosts. I set a policy to tolerate 1 failure. I pulled the network cables on one of the ESXi hosts. My VM was still not available after 30 minutes. Shouldn’t it still be available?
A: Yes, If HA is enabled. Please ensure HA is enabled so the VM can start elsewhere in the cluster.
Q: Is there any way to stop a resync operation, or halt it? During VM replication to VSAN, upon powering up VM on vSAN, storage policy takes effect and a resync operation starts.
A: There is no “recommended” way to stop resync this is an expected process. Interrupting it is unsupported at this time.
Q: Are there any recommendations or best practice to follow when you are migrating VMs from non-Cluster vSAN to the vSAN cluster?
A: If both clusters are part of the same vCenter this is a straight forward migration back and forth. Best practice is to test your environment first with non-critical and or test VMs first.
Q: What is most common issue on upgrades?
A: Depending on versions there can be a few different things. Ensure you have run all pre-checks. people performing upgrades before checks can be a big issue. Also ensuring you have enough cluster resources is huge as you need to ensure you can satisfy policies during upgrade.
Q: If I have both vSAN and HA enabled and I pull the NICS on an ESXi host, how long should the failover take?
A: HA is quite fast. It takes as long as is needed to register the VM on the new host and power it on. HA typically takes a matter of seconds to get the VM powered on, then however long the VM takes to boot. It is sometimes slower but HA is a quick process.
Q: What is the free space recommendable in vSAN?
A: 30% per disk group is what we would like so we can rebalance easily if that is required.
Q: Does it make sense, to install more than 1 vSCSI controller in a VM and attach multiple disks to different vSCSI controllers? Will this give me more IOPs?
A: To improve IOPS we usually recommend looking at more disk groups so you can take advantage of a larger amount of cache.
Q: What are the logs we need to collect when there is a problem?
A: Mostly clomd, vobd.log and command outputs help a lot. More details in this KB article.
