PowerProtect: TSDM backups fail with ABV0016 against a specific ESXi host
Summary: The virtual machine (VM) Protection Policy encounters intermittent backup failures during the scheduled execution with critical message ID ABV0016.
Symptoms
ABV0016: VM Direct engine 'localhost' is unable to back up the virtual machine 'my-vm-name' on vCenter 'my-vcernter-name' because of a vCenter API issue. The VM Direct engine cannot complete the backup of this virtual machine because of a vCenter API operation did not complete successfully. To troubleshoot this issue: 1) Verify that the vCenter can be reached. 2) If the vCenter is busy, wait for some activities to complete and then retry the operation. 3) Verify that the virtual machine exists on the vCenter. Read Less Unable to create LWD snapshot. vCenter task 'task-#####' failed: VimFault: dp.vpx.fault.DpdThrottleLimitExceeded.summary'
YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: SDM Full Sync: Closed file 'VM_Image_Daily_NON-SQL-uklpdtsbk001-f4723/PLCTLP-2832693c-377a-4f7e-952f-ffc1fdade2b1/Backups/c57b4143-f4c3-4760-86f5-7448429205ae/501efcd4-1f7e-5c9d-b98c-43633d01aa4c/lastSdmDiskBackupPath.json'. YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: SDM Full Sync: Checking if the last path exists which was used as a baseline for each disk ... YYYY-MM-DD HH:MM:SS INFO: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: SDM Full Sync: Full sync is not needed, skipping. YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: vCenter SOAP call 'DpSnapshot' was accepted. YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: The vCenter task failed: dp.vpx.fault.DpdThrottleLimitExceeded.summary YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: The vCenter task for 'Create SDM Snapshot' completed with state 'error'. YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: Unable to create SDM snapshot. YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: Retrying in 500ms... ... YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: vCenter SOAP call 'DpSnapshot' was accepted. YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: The vCenter task failed: dp.vpx.fault.DpdThrottleLimitExceeded.summary YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: The vCenter task for 'Create SDM Snapshot' completed with state 'error'. YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: Unable to create SDM snapshot. YYYY-MM-DD HH:MM:SS ERROR: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Create LWD Snapshot: Unsuccessful after 98 attempts over 1m0.158271936s. YYYY-MM-DD HH:MM:SS TRACE: [a30a8b8946958a4e;8d8911578980c3ff] SDM Data Mover: Snapshot failure can't be remedied by a Full sync.The vSphere web client shows at least one ESXi host with continuous LWD-based errors during the scheduled execution.
The vCenter Support bundle /var/log/vmware/vsan-health/vsanvcmgmtd-*.log output shows the ESXi host is being identified as a legacy DPD host:
YYYY-MM-DD HH:MM:SS warning vsanvcmgmtd[11593] [vSAN@6876 sub=dps] [opId=0849c23c] Found legacy DPD on host: host-29. Applying legacy throttle limit: '10', instead of regular throttle limit: '20'
Cause
If the vCenter vSAN Health Service cache has outdated version information or a null entry, it categorizes the ESXi host as legacy. The default PowerProtect Data Manager throttle sends 18 SDM backup sessions per ESXi host. The ESXi host processes the initial 10 SDM backups, but the vCenter would fail any subsequent SDM backups indicating the DPD limit was exceeded.
Resolution
There are enhancements in the vSphere 8.0 U1 that automatically refresh the vSAN Health service cache in certain conditions. It is recommended to upgrade to vSphere 8.0 U1 to take advantage of these enhancements and prevent certain vSAN Health service cache symptoms.
VMware Workaround:
Ensure the vCenter server and all the ESXi hosts in the cluster are at version 7.0 U3d or newer. Clear the cache by restarting the vCenter Server Appliance (VCSA) vSAN Health service over an SSH session:
service-control --restart vsan-health
If required, contact VMware Support for further insight or assistance with the vSAN Health service.
Additional Information
PowerProtect Data Manager Workaround:
There is a method in PowerProtect Data Manager to decrease the throttle value for TSDM backup sessions against all ESXi hosts. This can be used as a temporary workaround while upgrading the ESXi hosts or working with VMware Support.
- Ensure that no protection jobs are running
- Log in to PowerProtect Data Manager and elevate to sudo: sudo su -
- Make a copy of the vmdirect.env file: cp vmdirect.env vmdirect.env.orig
- Open environment vmdirect.env file: vi /opt/emc/vmdirect/unit/vmdirect.env
- Change the value for the environment variable "MAX_SDM_BACKUP_SESSIONS_PER_HOST" to 10 (instead of 18) and save the file.
- Restart the VISD service:
systemctl restart visd.service
Example output:
my-ppdm:/opt/emc/vmdirect/unit # cp vmdirect.env vmdirect.env.orig
my-ppdm:/opt/emc/vmdirect/unit # ls -l
total 20
-rwxr-xr-x 1 root root 397 Nov 2 22:48 visd.service
-rwxr-xr-x 1 root root 2225 Nov 2 22:48 vmdirect.env
-rwx------ 1 root root 2225 Apr 21 09:39 vmdirect.env.orig
-rwxr-xr-x 1 root root 360 Nov 2 22:48 vpod-nas.service
-rwxr-xr-x 1 root root 334 Nov 2 22:48 vpod.service
my-ppdm:/opt/emc/vmdirect/unit # vi vmdirect.env
my-ppdm:/opt/emc/vmdirect/unit # cat vmdirect.env | grep SDM
SDM_ENABLED_EXTERNAL_PROXIES=false
export SDM_ENABLED_EXTERNAL_PROXIES
MAX_SDM_SESSIONS_PER_HOST=20
export MAX_SDM_SESSIONS_PER_HOST
MAX_SDM_BACKUP_SESSIONS_PER_HOST=10
export MAX_SDM_BACKUP_SESSIONS_PER_HOST
MAX_ESX_SDM_TOTAL_DISKS=40
export MAX_ESX_SDM_TOTAL_DISKS
my-ppdm:/opt/emc/vmdirect/unit # systemctl restart visd.service
my-ppdm:/opt/emc/vmdirect/unit # systemctl status visd.service
● visd.service - VMDirect Infrastructure Services Daemon
Loaded: loaded (/usr/lib/systemd/system/visd.service; enabled; vendor preset: disabled)
Active: active (running) since Fri YYYY-04-21 09:42:38 PDT; 7s ago
Main PID: 67488 (visd)
Tasks: 16 (limit: 512)
Memory: 23.2M
CPU: 317ms
CGroup: /system.slice/visd.service
└─67488 /opt/emc/vmdirect/bin/visd -user admin -group app -p 9097 -nodemgmt -proxymgmt -vmdeployment -installclient -l
Apr 21 09:42:38 my-ppdm.my-domain.com systemd[1]: Started VMDirect Infrastructure Services Daemon.
Apr 21 09:42:38 my-ppdm.my-domain.com visd[67488]: INFO: Found GID 2000 (app) for group 'app'.
Apr 21 09:42:38 my-ppdm.my-domain.com visd[67488]: NOTICE: Effective group ID is set to 2000 (app).
Apr 21 09:42:38 my-ppdm.my-domain.com visd[67488]: INFO: Found UID 500 (admin) for user 'admin'.
Apr 21 09:42:38 my-ppdm.my-domain.com visd[67488]: NOTICE: Effective user ID is set to 500 (admin).
my-ppdm:/opt/emc/vmdirect/unit #