VMs repeatedly corrupted

JJordan354 · Wed Jan 07, 2026 12:20 pm

Hello!

I have noticed a pattern of VM corruption on one of my two-node Starwind vSAN clusters for the past two years. Oddly, the VM behavior described in this post is nearly identical to what I am seeing. viewtopic.php?t=5320

Starwind version: 8.0.14120

Any guidance on what could be causing this would be appreciated.

Wed Jan 07, 2026 12:32 pm

Did you try removing the cache?
Also, the system from the thread you were referring to had major flaws.

JJordan354 · Wed Jan 07, 2026 12:46 pm

Forgive me, but I don't see any obvious way of disabling this in the UI? FWIW, I see cache mode as "N/A" on each CSV.

However, I do see "Flash Cache" is enabled. I don't see any obvious was of disabling this either.

Wed Jan 07, 2026 1:46 pm

https://knowledgebase.starwindsoftware. ... -l1-cache/ - RAM Cache
https://knowledgebase.starwindsoftware. ... dance/661/ -that's for SSD cache

JJordan354 · Wed Jan 07, 2026 2:02 pm

Appreciate the guidance here. Just a bit of background, I have had this two-node cluster (configured by Starwind support originally) for about six years. What am I gaining/losing by disabling caching? Are there other underlying factors to look at before disabling a working configuration?

Regarding the post I referenced, I was just referring to the VM behavior only. That is, I build a VM and it works perfectly fine for a period of time and then suddenly becomes unstable/corrupted. Sometimes it days or weeks before I notice the behavior.

Wed Jan 07, 2026 3:22 pm

Just make sure to follow the regular restart precautions while working on it.
There might be a slight performance impact after disabling. Write-back caching is what boosts writes. Disabling it will affect the performance a bit.
Ideally, a configuration check could help shed some light, too, but it's possible only for the users who hold the support contract.

JJordan354 · Wed Jan 07, 2026 4:13 pm

Yeah, I dropped support for this particular cluster, while I do have support for two other clusters.

To be clear on the articles you provide, you want me to disable caching via removing these two header groups (Caching & Storages) from the configuration on both nodes?
<caching>
<cache type="write-through" size="430" units="GB" level="2">
<storages>
<storage_ref id="1"/>
</storages>
</cache>
</caching>
</storage>
</storages>
</device>
<system>
<resources>
<storages>
<storage id="1" name="My Computer\E\Flash-CSV\Flash-CSV.swdsk" type="device" lun="0x0">
<interval size="430" units="GB"/>
</storage>
<storage id="2" name="My Computer\D\CSV\CSV.img" type="file">
<interval size="3548" units="GB"/>
</storage>
</storages>
<network/>
</resources>

Wed Jan 07, 2026 6:13 pm

Yes. You need to remove caching and the corresponding storage areas.
You will also need to delete the corresponding device from StarWind.cfg on each node.
Make sure to follow the general restart precautions before stopping the service, and have a copy of the config files ready.

Again, I am not sure if the system was ever reviewed, and caching removal might not be what fixes the issue.