Azure-docs: Clarity on temporary disk

[dcbrown16]

The section on the temporary disk links to this blog:
https://blogs.msdn.microsoft.com/mast/2013/12/06/understanding-the-temporary-drive-on-windows-azure-virtual-machines/
The info there states that temporary disk data may be lost
"When you resize the VM, When you shutdown or restart your VM, when your VM is moved to a different host server(due to service healing, shutdown and restart), when the host is updated, when the host experiences a hardware failure etc."
However this page currently states "During a successful standard reboot of the VM, the data on the temporary drive will persist."

My understanding is that temporary disk data persists through a guest OS restart, but may be lost through a portal restart or putting the VM in Stopped (Deallocated) from the portal. Can the specifics here be clarified?
Having the old blog as reference is better than nothing, but it would be ideal to have the information replicated here or on another page instead so it can be localized and kept up to date.

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 3799662b-8f0f-c4e6-26b5-b2ff24a9242a
• Version Independent ID: 0c93223c-f28b-7ca7-0d10-c19e5fea9197
• Content: About unmanaged (page blobs) and managed disks storage for Microsoft Azure Windows VMs
• Content Source: articles/virtual-machines/windows/about-disks-and-vhds.md
• Service: virtual-machines-windows
• GitHub Login: @roygara
• Microsoft Alias: rogarana

[jakaruna-MSFT]

Thanks for the question! We are investigating and will update you shortly.

[DiegoRamirez-MSFT]

hello Dillon

you are correct, basically the content of the temp drive will be lost in the event of the VM moving to a different node, when you reboot the OS that is not considered a de-allocating operation, as opposed to doing it from the portal (reboot or shutdown) which could make the VM land on a different node.
I agree with your feedback, perhaps that note could be updated so its clear from the first time you take a look at the document, I am assigning this to the author to make the require edits

thanks for your feedback

[dcbrown16]

Thanks @DiegoRamirez-MSFT! I think minimally what we need is:

• add a sentence saying that the portal Restart and Stop operations may lead to loss of the temporary drive data

• replace the confusing "successful standard reboot" phrase with something that clarifies that this refers to a guest OS restart from within the VM

@roygara let me know if you have any questions, thanks!

[roygara]

@dcbrown16 the article actually used to have information which covered this. It looks like it was removed. I'm investigating why this was done so before I take any further steps.

Thanks for making us aware that this happened.

[dcbrown16]

Great, thanks for looking into that.

[roygara]

@dcbrown16 I was unable to get any solid answers as to why, so I'm just going to re-add what was cut from the section. Which was: "During a standard reboot of the VM, the data on the temporary drive should persist. However, there are cases where the data may not persist, such as moving to a new host. Accordingly, any data on the temp drive should not be data that is critical to the system. "

[dcbrown16]

Thanks @roygara, what do you think about changing "standard reboot" to "reboot from within the guest OS" or something similar? I don't really know what "standard reboot" means but it could easily be taken to refer to a portal restart, which in my understanding may erase the data.
If that is one of the points you were unable to confirm, then I guess we'll have to go with what you have, but I have specifically had a customer confused by the term "standard reboot" (in Japanese, but the translation is fine).

Also FYI, I believe pricing pages are written by different people but this is what the Linux VM page has at the moment, just for comparison.
"A temporary disk is storage that’s attached directly to the deployed VM. Data on a temporary disk is lost once the VM is shut down."
https://azure.microsoft.com/en-us/pricing/details/virtual-machines/linux/#Linux

And if I can ask one more thing, is there a policy against updating the "last updated" date outside of major revisions? I see this frequently across docs, but the page in question here has been updated many times since Nov 2017. From our perspective, having that updated can improve customer confidence in the information presented.

[roygara]

@dcbrown16 I wasn't able to confirm that yet but I still have asks out, regarding the first point.

The pricing page is indeed written by different people and we have no direct control over what is there.

To the best of my knowledge our policy on that piece of metadata is not to update it unless there's been a significant revision or the article has been reviewed in its entirety and deemed current. This requires an SME. We share the same perspective you do and also understand the importance of freshness, it's just that one of those two things need to happen for such an update to occur.

[roygara]

@dcbrown16 I've finally got a reply. Apparently in this case standard reboot here refers to both a portal restart or a reboot from within the guest OS. Because of that, I think it makes sense to leave it as is.

Thanks again for bringing this to our attention!

please-close

[dcbrown16]

Thanks @roygara, but the fact remains that this section points to a blog for more information and the blog states that temp disk is lost "When you shutdown or restart your VM," among other things. If we have reliable newer information, I would greatly appreciate it if we can delete the blog reference and salvage any information there that isn't currently duplicated. Having officially approved and contradicting information makes document guidance tough.

[roygara]

@dcbrown16 Ah, fair point. I'll cut out the blog for now and we'll see about incorporating remaining bits later. I'll create a backlog item to that effect. Thanks for bringing that up again!