DMCA
Consistent Detection of Global Predicates (1991)
Venue: | In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging |
Citations: | 165 - 3 self |
Citations
2868 | Time, clocks and ordering of events in distributed systems
- Lamport
- 1978
(Show Context)
Citation Context ...ide different ways to manage the uncertainty of debugging in a distributed environment. The distinction between possibly and definitely arises from the relative nature of time in a distributed system =-=[4]-=-. To an omniscient observer with a global clock, two events that we regard as concurrent (i.e. determine this total order in any practical way, but determining that some predicate was possibly true pr... |
1208 | Distributed snapshots: Determining global states of distributed systems.
- Chandy, Lamport
- 1985
(Show Context)
Citation Context ...gram is ¢ ¢ ¢ ¢s¡ ¢¤£ � 3 : ¤ 11 � 3 : ¤ 9 � 1 : ¤ 9s¡ ¥ (at least two of the three � variables are greater than 10). halted in order to examine local state. This is so if the predicate is not stable =-=[1]-=-. Thus one must either record relevant program state and include it in the messages sent to the monitoring process, or rely on some program replay or reversible execution technique to recover the stat... |
743 | Virtual time and global states of distributed systems
- Mattern
- 1989
(Show Context)
Citation Context ...local event changes some local state on which Φ depends. We say that such events potentially change Φ. Then, ¨ 0 constructs the partial order of events, represented compactly as vectors of timestamps =-=[7]-=-, in order to determine whether possibly Φ or definitely Φ held. The first two conditions are illustrated in Figure 1. In this space-time diagram, three processes ¨ ¨ ¨ 3 have variables � 1� � 2� � 3 ... |
108 | Techniques for debugging parallel programs with flowback analysis.
- Choi, Miller, et al.
- 1991
(Show Context)
Citation Context ...¡¡ ¢¦£ � Currently � 1 � � 2 � 5). £ £ £ £ £ £ � 1 : ¤ 1 £ £ £ £ £ £ £ ¥ detecting anomalies operate. Recent algorithms for dynamically detecting access anomalies have also employed vector timestamps =-=[2,3]-=-. The Flowback Analysis technique [2], computes before and after vectors for each synchronization event in a program execution. These are directly related to ��� our minimum and � � ��� � � ����� � ma... |
44 | Detection of Global State Predicates
- Marzullo, Neiger
- 1991
(Show Context)
Citation Context ...d Φ. Another technique is for ¨�� to send an empty message to all other processes when ¨�� potentially changes Φ. These, and other techniques for limiting the number of global states are discussed in =-=[5]-=-. 2.2 Detecting Currently Φ In contrast to the previous two algorithms, detecting Currently Φ is computationally cheap but may block the monitored program, as discussed in Section 1. Figures 5 and 6 g... |
34 |
Anomaly Detection in Concurrent Software by Static Flow Analysis
- N, Osterweil
(Show Context)
Citation Context ...2, yet in fact the condition � Definitely � 1 2 � holds. Our notion of possibly and definitely is closely related to the problem of access anomaly detection in parallel debugging and program analysis =-=[10]-=-. An important difference is that we wish to detect predicates over the states of the processes, while the access anomaly conditions are predicates over sequences of states (typically these are the se... |
17 |
Simultaneous regions: A framework for the consistent monitoring of distributed systems
- Spezialetti, Kearns
- 1989
(Show Context)
Citation Context ...hat they can miss conditions that in fact held. For example, Figure 8 shows such an execution. If the messages in this figure correspond to the messages generated in establishing simultaneous regions =-=[9]-=-, then her protocol will not ¤ detect ¤ � 1 2, yet in fact the condition � Definitely � 1 2 � holds. Our notion of possibly and definitely is closely related to the problem of access anomaly detection... |
13 | Tools for Monitoring and Controlling Distributed Applications
- Marzullo, Cooper, et al.
- 1991
(Show Context)
Citation Context ...hat might change the predicate being detected. Of course the disadvantage is the the predicate must be specified before execution of the program. 3 Breakpoints and Tracepoints in Meta The Meta system =-=[6]-=- is a toolkit that provides the basic primitives needed to build a non-real-time reactive system. It finds application in areas such as distributed application management, performance monitoring, load... |
7 |
A generalized approach to monitoring distributed computations for event occurrences
- Spezialetti
- 1989
(Show Context)
Citation Context ...itor therefore can � consider 2� � the state 1 2 2 before Φ could become false. 2.3 Comparison with Other Algorithms The work most similar in spirit to ours are the protocols developed by Spezialetti =-=[8]-=-. In particular, her event holding condition is the same specification as our protocol for detecting Currently Φ, and the specification of her event occurrence condition is similar to the specificatio... |
4 |
The Task Recycling Technique for Detecting Access Anomalies OnThe -Fly
- Dinning, Schonberg
- 1990
(Show Context)
Citation Context ...¡¡ ¢¦£ � Currently � 1 � � 2 � 5). £ £ £ £ £ £ � 1 : ¤ 1 £ £ £ £ £ £ £ ¥ detecting anomalies operate. Recent algorithms for dynamically detecting access anomalies have also employed vector timestamps =-=[2,3]-=-. The Flowback Analysis technique [2], computes before and after vectors for each synchronization event in a program execution. These are directly related to ��� our minimum and � � ��� � � ����� � ma... |