VMs disconnecting randomly from the Network on Multiple Hosts
search cancel

VMs disconnecting randomly from the Network on Multiple Hosts

book

Article ID: 418739

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • VMs are getting disconnected from the Network and getting the below error.


  • ESXi - /var/run/log/vobd.log shows the errors below:
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3477497752498us: [vob.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Non-empty delta disk being opened
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3477542166589us: [esx.audit.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Non-empty delta disk being opened
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3481376185290us: [vob.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Bloom Filter full
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3481420649402us: [esx.audit.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Bloom Filter full
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3481673146918us: [vob.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Non-empty delta disk being opened
    [YYYY-MM-DDTHH:MM:SS]: [vmfsCorrelator] 3481717614779us: [esx.audit.vmfs.sesparse.bloomfilter.disabled] Read IO performance maybe impacted for disk vmname-flat.vmdk: Non-empty delta disk being opened

  • VM - /vmfs/volumes/<datastore_name_or_id>/<vm_name>/vmware.log shows the errors below:
    YYYY-MM-DDTHH:MM:SS]| vmx| | I005: SnapshotVMX_TakeSnapshot start: 'VEEAM BACKUP TEMPORARY SNAPSHOT', deviceState=0, lazy=0, quiesced=0, forceNative=0, tryNative=1, saveAllocMaps=0
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | I005: HBACommon: First write on scsi3:0.fileName='/vmfs/volumes/<datastore_id>/vmname/vmname-000002.vmdk'
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | I005: DDB: "longContentID" = "a224b7d96b238399216eeb33c3e01a7a" (was "f943a40bcc86c18a1ba59d1c3e63e024")
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | I005: DISKLIB-CHAIN : DiskChainUpdateContentID: old=0x3e63e024, new=0xc3e01a7a (a224b7d96b238399216eeb33c3e01a7a)
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | A000: <<< file Throttled >>>
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | I005: HBACommon: First write on scsi3:1.fileName='/vmfs/volumes/<datastore_id>/vmname/vmname_1-000001.vmdk'
    [YYYY-MM-DDTHH:MM:SS]| vcpu-0| | I005: DDB: "longContentID" = "e08eea36ebfc3edc68f2174adfef8c86" (was "e961c662885d85b5f558067e6968c37b")

Note: The preceding log excerpts are only examples. Date, time and environmental variables may vary depending on your environment.

Environment

VMware vSphere ESXi 7.0.x

Resolution

Workaround:

To workaround this issue without powering off the VM

  1. Disable the bloom filter on the ESXi host running the SQL VM by setting the config option.
    esxcli system settings advanced set -i 0 -o /SE/BFEnabled

  2. vMotion the VM to another ESXi host and back to the original ESXi host for the changes to take effect.

Note: In VMware vSphere ESXi 7.0 U3f and 8.0, the bloom filter is disabled by default. The workaround does not need to be performed manually

Additional Information

Impact/Risks:
Requires a power off and power on of the VM. Bloom filter was introduced to boost read IO performance in SESparse. When bloom filter is turned off, read I/O performance will be in-line.