You need to enable JavaScript to run this app.
Volcano Engine Function as a Service

Volcano Engine Function as a Service

Copy page
Download PDF
Resource management
Managing GPU function resources
Copy page
Download PDF
Managing GPU function resources
After the initial release of a function, you can go to the function details page and click the Resource management tab. Here, you can view the resource allocation and instance-level information of the function, and perform operations and maintenance on its instance resources.​
Maximum resource​
Displays the various resource limits for the GPU function, including GPU computing power, GPU video memory, memory specification, maximum memory resources, maximum graphics memory, and the instance count range. ​
  • Maximum memory = Memory specification × Maximum instance count​
  • Maximum graphics memory = Graphics memory specification × Maximum instance count​
  • Instance count range: This refers to the range from the minimum instance count to the maximum instance count.​
  • Minimum instance count: This refers to the number of reserved instances. The system will maintain at least this number of continuously running instances until you set it to 0. Reserved instances are billed continuously while they are running. For billing details, see Product billing.​
  • Maximum instance count: This is the maximum number of instances the function can scale to. You can adjust the maximum instance count to dynamically modify the resource limits of the function. We recommend setting the maximum instance count based on your business needs to prevent high costs caused by excessive function invocations.​
tip​
  • The maximum instance count applies per reversion. For example, if you set the maximum instance count to 2, both the online and canary reversions support up to 2 instances each.​
  • Adjusting the maximum instance count is unavailable when a canary reversion of the function exists.​
Usage ratio​
Displays the memory usage and graphics memory usage of the GPU function, as well as the remaining available memory and graphics memory capacity within the account. The statistics allow you to monitor the resource usage of this function.​
To apply for more resources, click Apply for resources.​
Instances​
Displays information for all instances of the function. Here, you can perform instance management operations, such as viewing instance logs, logging in to an instance for command-line operations, and migrating instances.​
View instance logs​
Instance logs are the standard output logs of an instance, including instance startup logs, error logs, and business logs printed by the code. Compared to persistently stored function logs, instance logs deliver better timeliness and are suitable for scenarios such as function debugging.​
tip​
If a scale-in removes an instance, the logs of the instance can no longer be viewed.​
Select the target function instance and click Logs in the Actions column to view the real-time logs of the function instance.​
Webshell​
With Webshell, you can log in to a function instance and run commands. For example, view the environment information of the instance and debug the function.​
warning​
  • Operate with caution after logging in to an instance. Your operations may affect live requests, and any resulting impact is excluded from SLA guarantees.​
  • A single connection can last for up to 15 minutes and will automatically disconnect upon timeout. If you need to continue using it, log in again.​
  • The runtimes provided by veFaaS have common Webshell tools installed by default. If you deploy functions using a container image, ensure that any necessary Webshell tools are packaged into your image.​
  • You can log in to only running function instances. If the instance is removed due to a scale-in, the connection will be automatically terminated.​
Select the target instance and click Webshell in the Actions column. After successful login, the page will redirect to the command-line interface of the instance, where you can run desired commands.​
Migrate instances​
If your function instance is abnormal, you can use the instance migration feature to create a new instance.​
  1. Select the target instance and click Migrate in the Actions column.​
  1. In the pop-up confirmation box, click Confirm. The platform will terminate the current instance and create a new instance.​
Last updated: 2026.08.19 14:06:06