ML Units exceed the allowed maximum - google-cloud-platform

When I'm trying to submit a job for training a model in Google Cloud-ML, I'm getting the below error.
RESOURCE_EXHAUSTED: Quota failure for project my_project.
The requested 16.536900000000003 ML Units exceed the allowed maximum of 15.To read more about Cloud ML Engine quota, see https://cloud.google.com/ml-engine/quotas.
- '#type': type.googleapis.com/google.rpc.QuotaFailure
violations:
- description: The requested 16.536900000000003 ML Units exceed the allowed maximum
of 15.
subject: my_project
Now my question is, will this quota reset after few hours or days? Or do I need to ask for an increase in ML Units? If so, how to do that?

lleo#'s answer is correct; this section states the following:
The typical project, when first using Cloud ML Engine is limited in
the number of concurrent processing resources:
Concurrent number of ML training units: 15.
But to directly answer your question: that is a per-job quota (not something that "refills" after a few hours or days). So the immediate solution is to simply submit a job with fewer workers.
Having access to limited resources with new projects is pretty common on Google Cloud Platform, but you can email cloudml-feedback#google.com to see if you're eligible for an increase.

As stated by Guoqing Xu in comments, there was an issue. Which is resolved from their side and I can submit my job successfully now.
For me, there was only one job and that too it's light. Moreover there are no parallel computations going on. Hence, I was puzzled why the quota has reached.
It's resolved now, working fine. Thanks Google team for resolving it. :)

You can read about default limits here here
You may ask for increasing the quota (on the API Manager page in the console). It looks like your issue is about concurrent ML units used, so you might either refactor your training pipeline or ask for a increased quota.

Related

RESOURCE_EXHAUSTED with Vertex Pipeline by leveraging the free trial on GCP

I am fairly new to GCP and I am playing around with it taking advantage of the free trial.
I would like to run this simple pipeline in Vertex from notebook, but once I run it, I get this error in the very first task.
com.google.cloud.ai.platform.common.errors.AiPlatformException: code=RESOURCE_EXHAUSTED, message=The following quota metrics exceed quota limits: aiplatform.googleapis.com/custom_model_training_cpus, cause=null;
I've looked at the quotas of the error and I have 1 CPU for each available region. Of course I can not edit them, because of the free trial.
I also made these other attempts without success:
Set the CPU limit equal to 1 on the pipeline component;
Use the less powerful machine available (n1-standard-4, which actually uses 4 vCPUs);
Run the pipeline in different regions;
Define and run the pipeline in a completely new project;
Define and run the AutoML pipeline for classification/regression, starting from the available models.
It seems rather strange to me that it is not possible to try this service with free trial, but I don't know how to solve the problem. Any ideas? Thanks
Which regions have you tried? You can check the regions available of the resource in question at the Quotas page within your project. go to "IAM & Admin" > "Quotas" then go find the resource at the search bar:
Another alternative is to request a quota increase, be aware though that there that quota increase request will go though evaluation before getting granted. For more information about conditions about quota you can visit Google's documentation here.

GCP Quota request denied for GPU (All Regions) after "Quota 'GPUS_ALL_REGIONS' exceeded. Limit: 0.0 globally."

I am attempting to learn the fast.ai course. This requires a GPU and I am trying to use GCP.
I am following this guide here: https://course.fast.ai/start_gcp.html
I attempted increasing the quota for GPU (All Regions) jus like the guide, by setting a limit of 1 but I got this email from the support:
"Unfortunately, we are unable to grant you additional quota at this time. If
this is a new project please wait 48h until you resubmit the request or
until your Billing account has additional history."
I asked them if this meant they could not increase my quota for GPU (All regions) and this was the reason:
"Nevertheless, after careful evaluation on your project _______ we have determined that we are unable to grant your quota increase due to insufficient service usage history within your preferred project.
We suggest for you to make use of your current quotas and/or other resources readily available to serve your purposes for the meantime. To discuss further options on higher quota eligibility, please reach out to your Sales team [1] and provide this case ID ______ as reference. You may also direct them to reach out to us for questions or clarifications about your request."
Does anyone know how to proceed? I appreciate any advice on this. Thank you!
Some regions have better GPU availability than other so have a look at the documentation and try moving (or create a new) VM - you may have more luck with this.
Unfortunatelly as #John Hanley said - there's nothing we can do if you were denied increased quotas. You try requesting increased quota for one region (not all) which should (IMHO) be more feasible.

Why getting “Resource has been exhausted” although the usage is within the quota limit?

The problem
I'm using Google Sheets API v4 from node.js (Cloud Functions for Firebase) via googleapis npm package to write to the Google Sheets.
Getting this error sporadically:
Error: Resource has been exhausted (e.g. check quota).
at Gaxios._request (/workspace/node_modules/gaxios/build/src/gaxios.js:89:23)
at process._tickCallback (internal/process/next_tick.js:68:7)
It doesn’t say which particular quota is reached, though.
At that particular moment, indeed, there were many concurrent function executions that write to the Google Sheets, and several of them failed:
GCP console
https://console.cloud.google.com/iam-admin/quotas says that the quota statuses are fine:
https://console.cloud.google.com/apis/api/sheets.googleapis.com/quotas shows that the usage at the moment of issue is within the quota limits:
The guess
I'm suspecting that I may reach the “Write requests per 100 seconds per user” quota limit… because the “traffic by response code” chart shows 2.03 reqs/sec, which gives 203 requests per 100 seconds, and all of them from one credential:
However “Per-user quota usage is not displayed”, and neither error message or GCP console give definite indication on what is the actual quota being exhausted.
Also, there is a counter-argument to this guess: total “requests per 100 seconds” in the chart above is below 100, and I presume that “requests per 100 seconds per user” cannot be higher than total.
The questions
Is my conclusion based on the charts in the GCP console correct that I’m not reaching the quota limits?
How to figure out which quota limit is referred to in the error message?
What other cause can be there behind Resource has been exhausted (e.g. check quota) error?
Update based on the comment
Here are the links to the actual code repository.
The cloud function that is triggered: the line that calls my repository class method.
The repository method itself that calls the Spreadsheet class method to update the rows.
The Spreadsheet method which prepares data for updating/appending.
Actual method that calls the googleapis.
This is not a Minimal, Reproducible Example, this is the actual code used. I don't think providing a reproducible example is worth time in this case, because the question is primarily related to interpreting the error message and figuring out which particular quota limit has been reached.

Cannot Extend GPU Quota on Google Cloud

I am using Google Cloud for development and training of deep neural networks. I've reached the limits of what I can do with CPUs and now need to create and instance with one or more GPUs.
I've followed the instructions from multiple sources. As the instance was being created I received a notification that my quota for my region (us-west1) was zero and to request an increase.
I did so and received the confirmation email within minutes. However, when I then attempted to recreate the instance I was again met with the quota increase error.
I submitted another request (same region) but heard nothing.
I tried in a different region, again requesting a quota increase, but heard nothing. I did this 6 times and -- as you might have guessed -- neither received a confirmation email nor was I able to create my instance.
I tried the hack of using Chrome in Incognito mode, but no joy.
This was an issue a few months ago, at least judging from the S/O and Google forum posts. I would think that by now it would be fixed.
Any help would be much appreciated as I'm totally stuck
NB: Cross-posted to the gce-discussion forum
I think you should contact the Google Cloud Platform Support for this kind of issues.
Open a case asking why your quota increase has not been applied and I am sure they are going to solve this in some days or at least to tell you why your request was declined.
Notice that quoting from the official Documentation "Free Trial accounts do not receive GPU quota by default."
Disclaimer: I work for the Google Cloud Support.

google prediction api pricing

For google prediction api in the documentation pages it'showing different quota limits in different places for 10$ plan.
In the above link its saying that the prediction limits are 10000/day
https://cloud.google.com/prediction/
whereas in the next link it's saying the limit is 10,000/month
https://cloud.google.com/prediction/pricing?csw=1
If there is anybody who has used this and could tell me which is the correct one I would really appreciate it.
I'd say it's a typo on the first link. 10K free per day seems way too high -
considering you need to contact Google if you're going to do more than 40K per day (see "Usage Limits).
For now, until a Googler can confirm, I'd go with per month i.e. https://cloud.google.com/prediction/pricing?csw=1