Using Azure openAI key rotation automation

题意:使用 Azure OpenAI 密钥轮换自动化

问题背景:

We are planning to do the Azure OpenAI key rotation automatically. How can we achieve this? Do we have terraform resource for this.

我们计划自动执行 Azure OpenAI 密钥轮换。我们如何实现这一点?是否有相应的 Terraform 资源可用?

python 复制代码
resource "azurerm_cognitive_account" "example" {
  name                = "xxxxx"
  location            = azurerm_resource_group.example.location
  resource_group_name = azurerm_resource_group.example.name
  kind                = "OpenAI"

  sku_name = "S0"
}

问题解决:

Firstly, there is no direct resource present for the Open AI key rotation in terraform. Following a workaround on the requirement, I found below approach to make it work.

首先,Terraform 中没有直接用于 OpenAI 密钥轮换的资源。针对这一需求,我找到了一种可行的替代方法,具体如下。

As I mentioned in the comments, use azurerm_cognitive_account resource from terraform by providing kind as an "OpenAI" as shown.

正如我在评论中提到的,可以使用 Terraform 中的 azurerm_cognitive_account 资源,并将 kind 设置为 OpenAI,如图所示。

I tried creating a new open AI account with the below code and the deployment was successful.

我尝试使用以下代码创建一个新的 OpenAI 帐户,部署成功。

python 复制代码
provider  "azurerm"{
  features{}
}
data  "azurerm_resource_group"  "example"  {
    name = "DefaultResourceGroup-EUS"
}
resource  "azurerm_cognitive_account"  "example"  {
    name = "examplesample"
    location = data.azurerm_resource_group.example.location
    resource_group_name = data.azurerm_resource_group.example.name
    kind = "OpenAI"
sku_name = "S0"
}
}

Once it is done, you need to retrieve the keys from the external path. To do that, use data "external" block.

完成后,你需要从外部路径检索密钥。为此,可以使用 data "external" 块。

How to use it: Reference

Sample data block shown below:

示例数据块如下所示:

python 复制代码
data "external" "keys" {
  program = ["sh", "/path/retrieve_sshkey.sh"]
}

Now refer this in the main.tf terraform resource with null_resource block by passing the open ai resource id under triggers block and add a provisioner as well.

现在,在 main.tf Terraform 资源中,通过在 null_resource 块的 triggers 块中传递 OpenAI 资源 ID,并添加一个 provisioner

python 复制代码
resource "null_resource" "samplerotation" {
  triggers = {
    open_ai_resource_id = azurerm_cognitive_account.example.id
    }
provisioner "remote-exec" {
 //write a powershell script here and refer the above keys data block here
 }
}

Alternatively, you can also follow an other approach with the help of key vault. Store all the keys in the key vault and apply the key rotation from there itself.

另外,你也可以使用密钥保管库(Key Vault)采取另一种方法。将所有密钥存储在密钥保管库中,并从那里执行密钥轮换。

To do so, refer the terraform code from SO & for CLI approach refer Github doc.

为此,可以参考来自 Stack Overflow 的 Terraform 代码,并且对于 CLI 方法,可以参考 GitHub 文档。

相关推荐
哥布林学者1 天前
深度学习进阶(三十一)FlashAttention:IO 感知的精确注意力
机器学习·ai
岳小哥AI1 天前
AI大模型"幻觉"从何而来?解密GPT-4、DeepSeek一本正经胡说八道的真相
ai·ai基础
JaguarJack2 天前
Openai Codex 重大更新 已支持接入任意开源大模型
ai·openai·codex
Artech2 天前
[MAF预定义的AIContextProvider-02]AgentSkillsProvider——将Agent Skills引入MAF
ai·c#·agent·agent skills·maf
岳小哥AI2 天前
读懂计算机视觉CV、语言感知(ASR/TTS)、多模态,就能理解AI是如何“看到”与“听到”世界的
ai·ai基础
大树883 天前
金刚石散热越强,管路越先见顶
大数据·运维·服务器·人工智能·ai
施小赞3 天前
普通 RAG vs GraphRAG 核心对比
人工智能·ai
goldenrolan3 天前
A公司物料替代测试系统 v1.7:从需求到 exe/apk 的 AI 辅助全链路实践
android·自动化测试·软件测试·python·ai
gis分享者3 天前
GPT-Image-2 图像生成模型新手实战指南
gpt·ai·image·模型·图像生成
ofoxcoding3 天前
在AI API聚合平台配置DeepSeek V3.2提示词缓存实战:快速接入与成本优化指南
人工智能·spring·缓存·ai