-
Notifications
You must be signed in to change notification settings - Fork 0
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Storage Docs / Notifications #116
base: main
Are you sure you want to change the base?
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Suggested some minor tweaks/clarifications but mostly LGTM!
@Joe-Heffer-Shef Anything else you'd suggest including? |
This is designed to ensure you've not accidentally left a machine running idle. Unlike on-premise VM's that run 24/7 we recommend you shut down your instances when not in use, much like you would your own PC. | ||
|
||
We do also understand this alert could be a false positive where your workloads are not CPU demanding but still require the machine be on for extended periods. | ||
If this is the case please get in touch and we can make the instance exempt from these alerts. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It's more likely that the user has provisioned a far-too-big machine or doesn't understand how to use parallel processing, so it might be worth suggesting that they spin up a smaller machine if appropriate for that workload.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm trying to avoid the discussion of 'right sizing' in this specific doc. I think it needs it's own page / paragraph to be referenced.
That said I'm not sure how to approach the topic as sizing is going to be very specific to the users needs and AWS' instance types are semi-regularly changing. It's hard not to just point to the AWS docs, which isn't helpful for a majority of people.
I fear the doc we make will end up being too simple. This is a topic we're eager to push over to Ronin to see if they can make the UX explain this better so that people are less likely to pick a silly instance for their workload.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
https://blog.ronin.cloud/selecting-machine/
Turns out they do have a doc for this. It's probably a good place to signpost people to.
Q: What's the likelyhood a user will acutally read this.
ronin/drive-storage.rst
Outdated
When creating a new non-root drive, Ronin gives you multiple options for drive types. | ||
As a general rule of thumb we recommend you select **SSD**, this is due to how AWS provisions drive speed. | ||
|
||
The SSD drives will be allocated 125MiB/s and 3000iops as per `gp3 defaults <https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/general-purpose.html#gp3-performance>`_. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is fine but a bit cryptic, agree with Will's proposal
|
||
If you have received an email titled "Unused Drive Storage Detected" this means that Ronin has noticed detached drives have been in your project for extended periods. | ||
|
||
This could be from a terminated instance that had the "Keep On Termination" flag set: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Describe what happens when this option is set. People won't necessarily understand what "keep on termination" means.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I try to avoid adding bits like that as I feel it becomes a fine line of teaching people to suck eggs. This is where I'm disappointed that Ronin don't have their own end user documentation we can reference...
Do we really have to write our own docs on how another companies product works?
@willfurnass Maybe we add that to my list of questions for Ronin...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
People won't understand these concepts already so it's better for it to be explained simply and clearly. (It will feel too simple for us but it will be useful for them to have it spelled out.)
But I agree, the RONIN docs need to be better, it's not our job to explain their product.
ronin/drive-storage.rst
Outdated
======================================= | ||
|
||
Drive storage or block storage as it's often referred as is the storage attached directly to your :term:`instances<instance>` within Ronin. | ||
These are most commonly the "Root Drive" however Ronin gives you the option to create your own additional storage to attach and move between instances. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Link to the relevant RONIN documentation so people can learn how to do this
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
https://blog.ronin.cloud/storage-help/
This?
We already link to a few docs from here https://docs.rcc.shef.ac.uk/ronin/index.html but It looks like it's worth adding.
bcab9e1 Adds the link in.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
As a general rule of thumb we recommend you select **SSD**, this is due to how AWS provisions drive speed. | ||
|
||
The SSD drives will be allocated 125MiB/s and 3000iops as per `gp3 defaults <https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/general-purpose.html#gp3-performance>`_. | ||
Should your workload require more performance please do get in touch. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'd like a guide on using the storage-optimised machines properly please 😸
Most of the data processing tasks for CURED are bottlenecked heavily by disk I/O so users being able to easily optimise this would save a lot of time
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We could signpost to tooling that helps people quantify IO perf (on Linux and Windows instances - iostat
and iotop
for Linux).
A (mermaid.js) flow chart for diagnosing IO perf issues could be useful.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes please
I've collected a few tips but I realise it's probably a tricky topic for newcomers to grasp. There's a very wide potential variety of experience in the users - some extremely basic concepts to explain or at least signpost to relevant training
My overall feedback is: this is good, but in general there needs to be some basic computing training---not necessarily here in this documentation, but somewhere for researchers to get the necessary skills to use RONIN properly. Maybe some tutorials/walk-throughs that show in detail the steps to achieve common tasks? This is my RONIN Ubuntu cheat sheet if that helps at all. |
Co-authored-by: Will Furnass <[email protected]>
Co-authored-by: Will Furnass <[email protected]>
Co-authored-by: Will Furnass <[email protected]>
…ld/Sheffield_RCC_Docs into storage-notifications
I admit I didn't see this until I'd already replied to a bunch of the other suggestions. |
Yes I think an introductory RCC course is a very good idea. Also, Norbi is starting to talk about a new "research computing 101" talk to cover fundamentals that should be relevant to all different areas. |
Adds docs on Ronin notifications that might be recieved & some basic guidance on understanding drive storage.