Partial failures occur frequently in cloud systems and can cause serious damage including inconsistency and data loss. Unfortunately, these failures are not well understood. Nor can they be effectively detected. In this paper, we first study 100 real-world partial failures from five mature systems to understand their characteristics. We find that these failures are caused by a variety of defects that require the unique conditions of the production environment to be triggered. Manually writing effective detectors to systematically detect such failures is both time-consuming and error-prone. We thus propose OmegaGen, a static analysis tool that automatically generates customized watchdogs for a given program by using a novel program reduction technique. We have successfully applied OmegaGen to six large distributed systems. In evaluating 22 real-world partial failure cases in these systems, the generated watchdogs can detect 20 cases with a median detection time of 4.2 seconds, and pinpoint the failure scope for 18 cases. The generated watchdogs also expose an unknown, confirmed partial failure bug in the latest version of ZooKeeper.
more »
« less
Live Forensics for HPC Systems: A Case Study on Distributed Storage Systems
Large-scale high-performance computing systems frequently experience a wide range of failure modes, such as reliability failures (e.g., hang or crash), and resource overload-related failures (e.g., congestion collapse), impacting systems and applications. Despite the adverse effects of these failures, current systems do not provide methodologies for proactively detecting, localizing, and diagnosing failures. We present Kaleidoscope, a near real-time failure detection and diagnosis framework, consisting of of hierarchical domain-guided machine learning models that identify the failing components, the corresponding failure mode, and point to the most likely cause indicative of the failure in near real-time (within one minute of failure occurrence). Kaleidoscope has been deployed on Blue Waters supercomputer and evaluated with more than two years of production telemetry data. Our evaluation shows that Kaleidoscope successfully localized 99.3% and pinpointed the root causes of 95.8% of 843 real-world production issues, with less than 0.01% runtime overhead.
more »
« less
- Award ID(s):
- 2029049
- NSF-PAR ID:
- 10293041
- Date Published:
- Journal Name:
- Proceedings of the International Conference for High-Performance Computing, Networking, Storage and Analysis (SC 2020)
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Short time-to-localize and time-to-fix for production bugs is extremely important for any 24x7 service-oriented application (SOA). Debugging buggy behavior in deployed applications is hard, as it requires careful reproduction of a similar environment and workload. Prior approaches for automatically reproducing production failures do not scale to large SOA systems. Our key insight is that for many failures in SOA systems (e.g., many semantic and performance bugs), a failure can automatically be reproduced solely by relaying network packets to replicas of suspect services, an insight that we validated through a manual study of 16 real bugs across five different systems. This paper presents Parikshan, an application monitoring framework that leverages user-space virtualization and network proxy technologies to provide a sandbox “debug” environment. In this “debug” environment, developers are free to attach debuggers and analysis tools without impacting performance or correctness of the production environment. In comparison to existing monitoring solutions that can slow down production applications, Parikshan allows application monitoring at significantly lower overhead.more » « less
-
null (Ed.)Large-scale cloud services deploy hundreds of configuration changes to production systems daily. At such velocity, con- figuration changes have inevitably become prevalent causes of production failures. Existing misconfiguration detection and configuration validation techniques only check configu- ration values. These techniques cannot detect common types of failure-inducing configuration changes, such as those that cause code to fail or those that violate hidden constraints. We present ctests, a new type of tests for detecting failure- inducing configuration changes to prevent production failures. The idea behind ctests is simple—connecting production sys- tem configurations to software tests so that configuration changes can be tested in the context of code affected by the changes. So, ctests can detect configuration changes that ex- pose dormant software bugs and diverse misconfigurations. We show how to generate ctests by transforming the many existing tests in mature systems. The key challenge that we address is the automated identification of test logic and oracles that can be reused in ctests. We generated thousands of ctests from the existing tests in five cloud systems. Our results show that ctests are effective in detecting failure-inducing configuration changes before deployment. We evaluate ctests on real-world failure-inducing configura- tion changes, injected misconfigurations, and deployed con- figuration files from public Docker images. Ctests effectively detect real-world failure-inducing configuration changes and misconfigurations in the deployed files.more » « less
-
The successful deployment of autonomous real-time systems is contingent on their ability to recover from performance degradation of sensors, actuators, and other electro-mechanical subsystems with low latency. In this article, we introduce ALERA, a novel framework for real-time control law adaptation in nonlinear control systems assisted by system state encodings that generate an error signal when the code properties are violated in the presence of failures. The fundamental contributions of this methodology are twofold—first, we show that the time-domain error signal contains perturbed system parameters’ diagnostic information that can be used for quick control law adaptation to failure conditions and second, this quick adaptation is performed via reinforcement learning algorithms that relearn the control law of the perturbed system from a starting condition dictated by the diagnostic information, thus achieving significantly faster recovery. The fast (up to 80X faster than traditional reinforcement learning paradigms) performance recovery enabled by ALERA is demonstrated on an inverted pendulum balancing problem, a brake-by-wire system, and a self-balancing robot.more » « less
-
In this paper, we i) analyze and classify real-world failures of Kubernetes (the most popular container orchestration system), ii) develop a framework to perform a fault/error injection campaign targeting the data store preserving the cluster state, and iii) compare results of our fault/error injection experiments with real-world failures, showing that our fault/error injections can recreate many real-world failure patterns. The paper aims to address the lack of studies on systematic analyses of Kubernetes failures to date. Our results show that even a single fault/error (e.g., a bit-flip) in the data stored can propagate, causing cluster-wide failures (3% of injections), service networking issues (4%), and service under/over provisioning (24%). Errors in the fields tracking dependencies between object caused 51% of such cluster-wide failures. We argue that controlled fault/error injection-based testing should be employed to proactively assess Kubernetes' resiliency and guide the design of failure mitigation strategies.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.5555/3433701.3433787
