Persistent scratch location configuration on NFS using a Cisco Nexus 1000v is empty after a reboot

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

When configuring the persistent scratch location on NFS using a Cisco Nexus 1000v, you experience these symptoms:

• The configuration is written to the /etc/vmware/locker.conf file, but the configuration is not persistent across reboots.
• Configured persistent scratch locations on NFS are not persistent.
• Changes to the persistent scratch locations on NFS are lost when the ESXi host is rebooted.
• While creating the scratch location, you see an entry in the hostd.log file, located at /var/log/, indicating that the configuration is registered:
  
  2014-06-03T13:14:36.245Z [FFB2BD20 verbose 'Hostsvc.ScratchOptionProvider'] Registered advanced option 'ScratchConfig.ConfiguredScratchLocation'
  2014-03-03T13:14:36.245Z [FFB2BD20 verbose 'Hostsvc.ScratchOptionProvider'] Registered advanced option 'ScratchConfig.CurrentScratchLocation'
   

Environment

VMware ESXi 5.0.x

Cause

This issue occurs because the Cisco N1v VEM (Virtual Ethernet Module) is loaded via services.sh after the initialization of the scratch configuration during jumpstart.

• On the ESXi host console, you see entries similar to:
  
  2014-04-01T10:18:50Z jumpstart: current bootstate: BOOT_STATE_VALID
  2014-04-01T10:18:50Z jumpstart: mount point '/vmfs/volumes/UUID/.locker-esxi' not restored, waiting 2 seconds
  2014-04-01T10:18:52Z jumpstart: mount point '/vmfs/volumes/49da4bf0-b8c87bb8/.locker-dcpesx00000225' not restored, waiting 4 seconds
  2014-04-01T10:18:56Z jumpstart: mount point '/vmfs/volumes/49da4bf0-b8c87bb8/.locker-dcpesx00000225' not restored, waiting 8 seconds
  2014-04-01T10:19:04Z jumpstart: mount point '/vmfs/volumes/49da4bf0-b8c87bb8/.locker-dcpesx00000225' not restored, waiting 16 seconds
  2014-04-01T10:19:20Z jumpstart: mount point '/vmfs/volumes/49da4bf0-b8c87bb8/.locker-dcpesx00000225' not restored, waiting 32 seconds
  2014-04-01T10:19:52Z jumpstart: mount point '/vmfs/volumes/49da4bf0-b8c87bb8/.locker-dcpesx00000225' not restored, waiting 64 seconds
  2014-04-01T10:20:01Z crond[16943]: crond: USER root pid 17190 cmd /sbin/hostd-probe
  2014-04-01T10:20:02Z syslog[17191]: starting hostd probing.
  2014-04-01T10:20:17Z syslog[17191]: hostd probing is done.
  2014-04-01T10:20:56Z jumpstart: mount point '/vmfs/volumes/UUID/.locker-esxi ' not restored before timeout, givingup
   
• The third-party Nexus 1000v n1k-vem vDS module required to mount the NFS volume is only loaded later:
  
  2014-04-01T10:21:10Z root: init Running n1k-vem start
   
• The NFS volumes are restored after the jumpstart, later in the boot process:
  
  2014-04-01T10:21:21.074Z cpu6:17187)NFS: 331: Restored connection to the server <IP>/64 mount point /vol/datastorename/folder, mounted as 49da4bf0-b8c87bb8-0000-000000000000 ("volume-name")

Resolution

This is an expected behaviour. As the third-party vDS module is loaded after the initial jumpstart, there is no way to bypass this issue.

 

To use the persistent scratch location on a network store, use either the Virtual Standard Switch (vSS) or a standard VMware Virtual Distributed Switch (vDS) in your configuration.

Additional Information

For related information about persistent scratch location for ESXi 4.x and 5.x, see Creating a persistent scratch location for ESXi 4.x and 5.x (1033696).

Creating a persistent scratch location for ESXi 4.x/5.x/6.x
ESXi ホストを再起動すると、スクラッチロケーションの設定が空になる
使用 Cisco Nexus 1000v 的 NFS 上的持久暂存位置配置在重新引导后为空