0

我将 Cloudify 2.7 与 OpenStack Icehouse 一起使用。

我为 Glassfish 开发了一个服务配方，将实例注册到 Apache 负载均衡器上。

start 事件执行一个glassfish_start.groovy运行以下命令的文件：

su - glassfish -c "asadmin --passwordfile /home/glassfish/passwd.gf --user $userAdmin start-domain $domain"
su - glassfish -c "asadmin --passwordfile /home/glassfish/passwd.gf --user $userAdmin deploy --contextroot \"/\" $warName"

启动事件结束后，我可以看到：

2015-07-14 09:59:15,678 glassfish.glassfish [1] INFO ....
[org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy (startDetection): arePortsFree http=8080
2015-07-14 09:59:15,680 glassfish.glassfish [1] INFO [org.cloudifysource.usm.UniversalServiceManagerBean] - Process liveness test passed
2015-07-14 09:59:15,680 glassfish.glassfish [1] INFO [org.cloudifysource.usm.UniversalServiceManagerBean] - Executing Process Locators!
2015-07-14 09:59:15,684 glassfish.glassfish [1] INFO [org.cloudifysource.usm.locator.DefaultProcessLocator] - Looking for actual process ID in process tree
2015-07-14 09:59:15,692 glassfish.glassfish [1] INFO [org.cloudifysource.usm.locator.DefaultProcessLocator] - Located process (6446): /usr/lib/jvm/java-8-oracle/jre/bin/java [java, -jar, /opt/glassfish4/bin/../glassfish/lib/client/appserver-cli.jar, --passwordfile, /home/glassfish/passwd.gf, --user, <user>, deploy, --contextroot, /, hello.war]
2015-07-14 09:59:15,693 glassfish.glassfish [1] INFO [org.cloudifysource.usm.UniversalServiceManagerBean] - Monitored processes: [6446]

2015-07-14 09:59:15,700 glassfish.glassfish [1] INFO [org.cloudifysource.usm.USMEventLogger.glassfish.glassfish] - glassfish.glassfish-1 POST_START invoked
2015-07-14 09:59:15,707 glassfish.glassfish [1] INFO [org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy: glassfish Post-start - useLoadBalancer = true - Adding Glassfish instance onto ApacheLB...
2015-07-14 09:59:15,721 glassfish.glassfish [1] WARNING [org.cloudifysource.utilitydomain.admin.TimedAdmin] - Admin object might expire prematurely! The specified timeout for waiting for PU was set to 180 SECONDS while the admin timeout is 120 seconds
2015-07-14 09:59:15,740 glassfish.glassfish [1] INFO [org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy: add-node onto ApacheLB ...
2015-07-14 09:59:15,742 glassfish.glassfish [1] INFO [org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy: ipAddress is <IP> ...
2015-07-14 09:59:15,745 glassfish.glassfish [1] INFO [org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy: Sto per aggiungere l'URL http://<IP>:8080 ad ApacheLB ...
2015-07-14 09:59:21,686 glassfish.glassfish [1] INFO [glassfish-stdout] -      [exec] Application deployed with name hello.
2015-07-14 09:59:21,687 glassfish.glassfish [1] INFO [glassfish-stdout] -      [exec] Command deploy executed successfully.
2015-07-14 09:59:21,687 glassfish.glassfish [1] INFO [glassfish-stdout] -      [exec] startOnUbuntu.sh: End
2015-07-14 09:59:21,687 glassfish.glassfish [1] INFO [glassfish-stdout] -      [exec] + echo 'startOnUbuntu.sh: End'
2015-07-14 09:59:23,007 glassfish.glassfish [1] INFO [org.cloudifysource.dsl.internal.BaseDslScript] - glassfish-service.groovy: glassfish Post-start ended
2015-07-14 09:59:23,008 glassfish.glassfish [1] INFO [org.cloudifysource.usm.USMEventLogger.glassfish.glassfish] - glassfish.glassfish-1 POST_START completed, duration: 7.3 seconds

2015-07-14 09:59:23,008 glassfish.glassfish [1] INFO [org.cloudifysource.usm.UniversalServiceManagerBean] - Starting async tasks
2015-07-14 09:59:25,011 glassfish.glassfish [1] INFO [org.cloudifysource.usm.USMLifecycleBean] - Stop detection - service has stopped!
2015-07-14 09:59:25,011 glassfish.glassfish [1] INFO [org.cloudifysource.usm.UniversalServiceManagerBean] - Detected death of underlying process
2015-07-14 09:59:25,011 glassfish.glassfish [1] INFO [org.cloudifysource.usm.USMEventLogger.glassfish.glassfish] - glassfish.glassfish-1 POST_STOP invoked

为什么它调用了停止检测？

此外，Cloudify 文档说明了这一点：

已知限制：如果您的启动脚本返回，Cloudify 会将您的服务实例视为已停止，因此会尝试再次配置它。要避免这种行为，请确保您的启动脚本永远休眠。

我应该怎么办？

4

1 回答 1

0

默认情况下，Cloudify 会监控 start 生命周期事件启动的进程。如果该进程终止，Cloudify 将认为这是服务实例的故障并尝试重新启动它。

如果 start 事件启动的进程不是您应该监控的进程，（可能是因为您将进程作为服务运行，或者使用 nohup），您可以将 locator 部分添加到配方中，这会告诉 Cloudify 在哪里查找受监控的过程。您还可以将定位器配置为根本不监视进程，如下所示：

locator {
       NO_PROCESS_LOCATORS
}

这里有一个类似的问题：cloudify 2.7 locator NO_PROCESS_LOCATORS

于 2015-07-14T12:42:30.000 回答