AppSync:当使用最大数量的 vCenter (VC) 连接时,后续连接请求将等待,并且在等待时可能会超时。
요약: 同时需要超过允许的 4 个 VC 连接。
증상
同时需要超过允许的 4 个 VC 连接。
通过在 AppSync 服务器日志中查找与以下内容类似的消息,可以观察到这一点:
->>> getURLConnection::https://vcenter-IP/sdk reached maximum connection limit. Wait for sometime to get a free connection
->>> EJB default - 1Unlocking entityID:https://vcenter-IP /sdk for update
->>> --getConnection::Waiting on BUSY connection for entity:https://vcenter-IP/sdk
如果等待 VC 连接的服务计划超时,则服务计划可能会卡在 “in-progress” 状态。
可以通过以下方式在 AppSync 服务器日志中观察到服务计划的超时情况:
WARN [Transaction Reaper] [com.arjuna.ats.arjuna] [] [] ARJUNA012117: TransactionReaper::check timeout for TX 0:ffff0ae651aa:1c135b84:60be4e22:1f2b2c in state RUN
WARN [Transaction Reaper Worker 0] [org.hibernate.engine.transaction.synchronization.internal.SynchronizationCallbackCoordinatorTrackingImpl] [] [] HHH000451: Transaction afterCompletion called by a background thread; delaying afterCompletion processing until the original thread can handle it. [status=4]
WARN [Transaction Reaper Worker 0] [com.arjuna.ats.arjuna] [] [] ARJUNA012121: TransactionReaper::doCancellations worker Thread[Transaction Reaper Worker 0,5,main] successfully canceled TX 0:ffff0ae651aa:1c135b84:60be4e22:1f2b05
WARN [Transaction Reaper Worker 0] [com.arjuna.ats.arjuna] [] [] ARJUNA012095: Abort of action id 0:ffff0ae651aa:1c135b84:60be4e22:1f2b2c invoked while multiple threads active within it.
WARN [Transaction Reaper Worker 0] [com.arjuna.ats.arjuna] [] [] ARJUNA012108: CheckedAction::check - atomic action 0:ffff0ae651aa:1c135b84:60be4e22:1f2b2c aborting with 1 threads active!
WARN [Transaction Reaper Worker 0] [com.arjuna.ats.arjuna] [] [] ARJUNA012121: TransactionReaper::doCancellations worker Thread[Transaction Reaper Worker 0,5,main] successfully canceled TX 0:ffff0ae651aa:1c135b84:60be4e22:1f2b2c
卡在“中”的服务计划的后续运行in-progress“失败,表明服务计划已在运行。
in-Progress“服务计划”,并允许它再次运行。
원인
해결
增加以秒为单位指定的事务超时可能会解决超时问题。
默认值为 5 分钟,如果需要,可以增量增加。下面的示例将超时增加到 20 分钟。
要增加事务超时,请在 standalone.xml,位于 C:\EMC\AppSync\jboss\standalone\configuration 在 AppSync 服务器上:
- 备份
standalone.xml文件。 - 编辑
standalone.xml归档并增加:
<coordinator-environment default-timeout="300"/>
<coordinator-environment default-timeout="1200"/>
- 重新启动三个 AppSync 服务器服务。
计划与 vCenter 相关的服务计划以减少与 vCenter 的同步交互可能会解决此问题,并消除增加事务超时的需要。