Changes between Version 49 and Version 50 of Check
- Timestamp:
- Oct 28, 2010, 7:18:52 PM (14 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
Check
v49 v50 363 363 To simulate node failure, the nodes can be simply unplugged from their power source or the service running on them can be suspended using the appropriate URL, such as [http://localhost:8000/babe-00ff-fe00-0001/service/temp/stop] 364 364 365 [[Image(sh-suspend service.png, center )]]365 [[Image(sh-suspend service.png, center, 300px)]] 366 366 367 367 … … 370 370 The Self-healing service becomes aware of the islands being monitored using a flexible and standard compliant xml file. For this demo, a sample config.xml file is provided: 371 371 372 [[Image(sh-config.png, center )]]372 [[Image(sh-config.png, center, 300px)]] 373 373 374 374 The main entities described in the config.xml are: … … 381 381 For testing purposes, we have also created a client application that requests temperature information from nodes running the appropriate service. The interval between requests has been arbitrarily been set at 2 seconds. In the screenshot, you may see that, when a node fails, the number of read values drops: 382 382 383 [[Image(sh-no service.png, center )]]383 [[Image(sh-no service.png, center, 600px)]] 384 384 385 385 === Demo 2: Self-healing inside an island === … … 387 387 In this second demo we will be starting the Self-healing service and repeat the node failure simulation. 388 388 389 At the service's startup, it loads the config file and recognizes the resources that it needs to monitor :390 391 [[Image(sh-service startup.png, center )]]392 393 We repeat the node failure several times. In the screenshot you may see that node failure was repeated three times, but each time the self-healing service starts the service on a node to fill-in for the missing information . Only once the node failure was so close to the client's request that the latter could only read 1 value. The other times, the nodes were up again in time for the client to read enough values:394 395 [[Image(sh-start node.png, center )]]389 At the service's startup, it loads the config file and recognizes the resources that it needs to monitor (lower window): 390 391 [[Image(sh-service startup.png, center, 600px)]] 392 393 We repeat the node failure several times. In the screenshot you may see that node failure was repeated three times, but each time the self-healing service starts the service on a node to fill-in for the missing information (lower window). Only once the node failure was so close to the client's request that the latter could only read 1 value (upper window). The other times, the nodes were up again in time for the client to read enough values: 394 395 [[Image(sh-start node.png, center, 600px)]] 396 396 397 397 === Demo 3: Self-healing across islands ===