VMware Cloud Community
sumit_vmw
Contributor
Contributor

vSAN Congestion alarm in vcenter

Hello ,

Whenever I am a taking a snapshot of volume by

openstack volume snapshot command  , I am seeing a  congestion alarm as long as snapshot is getting created.

Once the snapshot is  created, all alrms turn green and vsan healthy

pastedImage_0.png

Any clues to move forward with investigation ?

Regards

Sum

Tags (1)
Reply
0 Kudos
1 Reply
TheBobkin
Champion
Champion

Hello Sum,

Welcome to Communities.

So, the first two pieces of information needed to understand this are the type of congestion (e.g. ssd, log, mem, comp) and amount.

Is this a homelab and/or a relatively low-spec vSAN cluster? (e.g. small or low-performance/read-intensive Cache-tier devices, SATA devices throughout etc.).

I ask as it is relatively common in clusters/Disk-Groups on the lower end of things to have a spike of ssd-congestion during periods of relatively bursty workloads against single device (e.g. snapshot actions) - this is basically vSAN intentionally throttling IO to the Cache-tier SSD so that it can keep up with the workload.

Some general information on what congestion is and the different types:

VMware Knowledge Base

We can see from the tasks in your screenshot that you may have been Storage vMotioning data to this cluster at the same time - maybe you were putting too much load on this (e.g. storage migrating data is a straight-up write operation, other VM workloads are running also and then you are taking snapshots on VMs on top of this).

Another thing of note is the version of ESXi and vCenter in use here - why this is relevant is the warning and alert trigger values for congestion were changed in 6.7 U1 from 32/64 (yellow/red) to 200/220 (partially due to changes in how vSAN works but also as these were initially set too low and caused undue concern as the components in question are just busy and thus moderate levels of congestion are intentional and expected):

VMware Knowledge Base

Bob

Reply
0 Kudos