再会，
我是Nutanix的新手，最近购买了一个集群。它仅运行了大约30天，现在管理我的网络，我正在对其进行一些配置。我需要一些帮助，并收到一个错误消息。我将AHV用作3个节点簇上的高探。在此集群上，我正在运行5个基于Windows的服务器VM（不使用Hyper-V或VMware）。我遵循了《管理指南》中的说明，通过在服务器上启用VSS影子副本，然后在所有服务器上安装访客工具并创建保护域Async DR。我的配置正在工作，正在为我的域控制器和应用程序服务器创建快照。但是，当试图为我的文件服务器创建快照时，我会继续遇到以下错误。
“警告：VS Snapshot因快照（169035，1563300389081879，960）在FileServer保护的VM（S）FS-01上失败，因为Quiesiescting Guest guest VM（S）失败或计时。
影响：碰撞一致的快照，而不是应用程序一致的快照。
原因：由于内部错误，来宾无法呼应VM。
分辨率：查看来宾VM中的日志。如果VM无法拨号，请减少VM上的负载，然后重试。
节点ID：HM186S003223
块ID：19SM6J220007
块类型：NX-1065-G6
集群ID：169035
群集UUID：00058DD0-3C5E-5317-0000-00000002944B
群集名称：ntnx
集群版本：EL7.3释放-Euphrates-5.10.8.1 stable-9AC2CB13B645B9DF04EB85B85B0E091F1060EE27439
集群IPS：192.168.1.10 192.168.1.11 192.168.1.12
时间戳：1月22日星期三10:03:48 AST 2020”
我非常感谢您的协助解决这个问题。
此致，
凯文·赫拉曼（Kevon Heraman）
最好的答案杰里米
Are you still having an issue?
For successful application consistent\u00a0VM snapshots, the following need to happen:<\/p>
- Backup snapshot is triggered with app-consistent option enabled.<\/li>
- The CVM reaches out to the NGT service running on the VM via TCP\/IP to signal that VSS snapshot is needed.
- This requires TCP\/IP communication be possible both ways, but should not need DNS since the NGT service will inform Prism of the IP, and the NGT installation has cluster IP info. (NAT could be a problem)<\/li>
- The NGT service must, of course, be installed and running, and able to reach the CVM on port 2074.
(more detail on what ports are needed for full NGT function here<\/a>)\u00a0<\/li>- Communication uses a pre-shared key which is part of NGT installation, and an identifier which is unique to the VM. To have this work, NGT installation must be unique per VM using the \u201cmount iso\u201d option from Prism.
If cloning VMs, you can pre-install NGT and then mount the ISO again on the clone before powering on. The NGT service will\u00a0fetch updated identifier info during service start\u00a0if the NGT ISO is found in the VM\u2019s CDROM drive.<\/li> <\/ol><\/li>- The NGT service requests VSS Quiesce operation from the Windows OS.
- In a quiesce operation, all new changes to disk are held in hot-backup in memory on the VM until the snapshot is finished.<\/li>
- All pending changes to disk must finish before the snapshot can happen.<\/li>
- This requires sufficient memory on the VM to hold all\u00a0new changes long enough to complete the snapshot, otherwise application consistent snapshot will fail. If you\u2019re seeing intermittent failures, this is where to focus.<\/li>
- This process can be impacted by high workload, slower disk performance, hypervisor memory or CPU contention, VM memory or CPU contention, or any Windows VSS specific issue which prevents quiesce completion.
- Options for resolution include re-balancing workloads, adding resources at the host or VM, or adjusting scheduled jobs so that snapshots can run at lower-IO times.\u00a0<\/li> <\/ol><\/li> <\/ol><\/li>
- Once VSS signals back to NGT that quiesce is good, NGT service signals to Prism and then snapshot is taken. Prism then signals back to NGT, which relays to VSS, at which point new pending disk operations are allowed to flow to disk.<\/li> <\/ol>
Where this\u00a0error message indicates \u201can internal error\u201d I would actually be looking at VSS and the Nutanix Guest Tools service on the user VM itself. A different error should be seen if Prism cannot reach the NGT service on the VM. There is yet another error when NGT has not been enabled.\u00a0
The KB article \u201cTaking app-consistent (VSS) snapshots using NGT fails on Windows VMs<\/a>\u201d covers one scenario where the culprit is anti-virus software on the VM. The KB also gives some good general steps for exploring the issue with Event Viewer and the vssadmin command. These are often essential in identifying and resolving the issue. The important thing to look for in \u2018vssadmin list writers\u2019 and \u2018vssadmin list providers\u2019 is the last error state. If the last attempt was successful we\u2019ll see an indication of no error. If you just tried the backup and still see no error here, our problem is happening before VSS gets triggered.
The\u00a0article \u201cNutanix Guest Tools Troubleshooting Guide<\/a>\u201d provides further guidance on validating the Nutanix Guest Tools installation.\u00a0<\/p>","className":"post__content__best_answer"}">
- Communication uses a pre-shared key which is part of NGT installation, and an identifier which is unique to the VM. To have this work, NGT installation must be unique per VM using the \u201cmount iso\u201d option from Prism.
