One of the major advantages of cloud spot instances in cloud computing is to allow stakeholders to economically deploy their applications at much lower costs than that of other types of cloud instances. In exchange, spot instances are often exposed to revocations (i.e., terminations) by cloud providers. With spot instances becoming pervasive, terminations have become a part of the normal behavior of cloud-based applications; thus, these applications may be left in an incorrect state leading to certain bugs. Unfortunately, these applications are not designed or tested to deal with this behavior in the cloud environment, and as a result, the advantages of cloud spot instances could be significantly minimized or even entirely negated. We propose a novel solution to automatically find these bugs and locate their causes in the source code. We evaluate our solution using 10 popular open-source applications. The results show that our solution not only finds more instances and different types of these bugs compared to the random approach, but it also locates the causes of these bugs to help developers improve the design of the shutdown process and is more efficient in finding instances of these bugs since it interposes at the system call layer.

中文翻译：

---

T-BASIR：在云现货市场中查找基于云的应用程序的关闭错误

Cloud Spot 实例在云计算中的主要优势之一是允许利益相关者以比其他类型的云实例低得多的成本经济地部署他们的应用程序。作为交换，Spot 实例通常会被云提供商撤销（即终止）。随着 Spot 实例变得普遍，终止已成为基于云的应用程序正常行为的一部分；因此，这些应用程序可能会处于不正确的状态，从而导致某些错误。不幸的是，这些应用程序的设计或测试并未针对云环境中的这种行为进行设计或测试，因此，云点实例的优势可能会被显着降低甚至完全否定。我们提出了一种新颖的解决方案来自动查找这些错误并在源代码中定位它们的原因。我们使用 10 个流行的开源应用程序评估我们的解决方案。结果表明，与随机方法相比，我们的解决方案不仅发现了更多的这些错误的实例和不同类型的错误，而且还定位了这些错误的原因，帮助开发人员改进了关闭流程的设计，更有效地找到了这些错误的实例。这些错误是因为它介入系统调用层。