VMware Cloud Community
srodenburg
Expert
Expert

6.2 - Filesystem overhead is huge ??

Hello,

Reference: 6 node hybrid cluster, FTT = 1, SW = 1

Was just messing about in the new 6.2 Monitor and Health screens today and noticed it now says "Virtual disks" = 1.86 TB and "FIlesystem Overhead" = 4.17 TB

Say what ????

A few days ago, Filesystem overhead said it was around 1.9 TB. This makes perfect sense to me because everything is mirrored. All i recall doing since then is ordering a re-balance but that finished succesfully yesterday.

Am i seeing things? Why did my overhead explode? and with what?

Related question:  what is that "other" stuff? It's massive as well. I run with "SwapThickProvisionDisabled=1" as you can see (swap objects=only 508 megs)

Reply
0 Kudos
7 Replies
zdickinson
Expert
Expert

Good afternoon, not sure why the Overhead got so big.  I can help answer the "other" question.

Other: Capacity consumed by VM templates, unregistered VMs, standalone VMDKs not associated with VMs, manually created VSAN objects, manually created directories that store ISOs for example.

From:  VSAN 6.2 Part 7 - Capacity Views - CormacHogan.com

Thank you, Zach.

Reply
0 Kudos
srodenburg
Expert
Expert

My system is doing the funky chicken. I expanded a VM's disk with 200GB. As a result, it now says "Virtual disks = 1.41 TB" (schrunk) and "Filesystem Overhead = 4.59 TB" (grown).

What dark sourcery is this...

Reply
0 Kudos
elerium
Hot Shot
Hot Shot

That does look really wonky, from what I see in my test 6.2 environment, the mirrored FTT copy still counts as "Virtual disks" and not as filesystem overhead. In my tests, I have high checksum overhead (415GB of virtual disk but 318GB of checksum overhead). I remember reading before that every 4k block gets a 5 byte checksum, in terms of checksum overhead, my ratio seems really high. Not sure if that's normal but I only have 3 small test VMs in this environment, so not sure.

vsan-capacity.png

Reply
0 Kudos
srodenburg
Expert
Expert

Oh it used to look 50-50'ish in my environment too for a while (which is exactly what one should expect in a fully mirrored env.) but has gone completely cookoo since a couple of days.

Since then, my environment goes all over the place. With the exact same VM's, but after a rebalance-action today (took a couple of hours to complete), it now looks like this:  Virtual disks: 1.62 TB, Overhead: 4.41 TB.

I'm telling ya, VSAN 6.2 is just making this shit up. It just rambles out some numbers and changes them every couple of hours when I happen to take a look. All the while, the dataset, the number of VM's and their size, stays the same. So it should not change. At least not much anyway.

Swap Objects went form 500.8 MB last night to 36.14 GB now. And nothing changed. I mean really??

The developer who wrote this stuff should stay off the booze i tell ya. I cannot take this feature seriously anymore. I'll just ignore it. Just like the "Proactive Multicast Test" which still fails 100% of the time despite everything being configured correctly. Honestly, don't make stuff like this GA when it's not finished...

Reply
0 Kudos
CHogan
VMware Employee
VMware Employee

Please open service requests for the anomalies folks - obviously something is not right.

Please share the SR numbers - it will let the VMware folks on the communities to track the progress.

Thanks

http://cormachogan.com
Reply
0 Kudos
MillardJK
Enthusiast
Enthusiast

In my case, the two environments I have both have the issue, and neither have any support contracts--they're vExpert labs. SR isn't an option...

——
Jim Millard
Kansas City, MO USA
Reply
0 Kudos
ChrisJ615
Enthusiast
Enthusiast

Any updates to this mystery? File System Overhead is 41% of my current consumption.  Would love to know how to identify what files are considered file system overhead and how to investigate.

Reply
0 Kudos