Migrate VM without moving the Flash read cache

[EddiePhoenix]

I have a VM with a large number of VMDKs each with individual flash read cache settings. When i place its host into maintenance mode the VM starts to migrate then fails. The only fix is to restart vCenter services and hard bounce the host.  I have been able to replicate this issue with a test machine by adding many VMDKs each with vFRC configured and doing a drag and drop migration in the vCenter heavy client. However when i use the web client and manually select dropping the cache it works without issue.

log data

2018-02-01T00:40:48.371Z| Worker#3| I125: Migrate: Remote Log: Destination waited for 132.56 seconds.

2018-02-01T00:40:48.379Z| vcpu-0| I125: VMXVmdb_SetMigrationHostLogState: hostlog state transits to failure for migrate 'to' mid 1517445513843299

2018-02-01T00:40:48.396Z| vcpu-0| I125: MigrateSetStateFinished: type=1 new state=6

2018-02-01T00:40:48.396Z| vcpu-0| I125: MigrateSetState: Transitioning from state 3 to 6.

2018-02-01T00:40:48.396Z| vcpu-0| A100: ConfigDB: Setting config.readOnly = "FALSE"

2018-02-01T00:40:48.396Z| vcpu-0| I125: Migrate_SetFailureMsgList: switching to new log file.

2018-02-01T00:40:48.400Z| vcpu-0| I125: Migrate_SetFailureMsgList: Now in new log file.

2018-02-01T00:40:48.495Z| vcpu-0| I125: Migrate: Caching migration error message list:

2018-02-01T00:40:48.495Z| vcpu-0| I125: [msg.checkpoint.precopyfailure] Migration to host <ip.add.re.ss> failed with error Connection closed by remote host, possibly due to timeout (0xbad003f).

2018-02-01T00:40:48.495Z| vcpu-0| I125: [vob.vmotion.stream.keepalive.read.fail] vMotion migration [a3e007a:1517445513843299] failed to read stream keepalive: Connection closed by remote host, possibly due to timeout

2018-02-01T00:40:48.495Z| vcpu-0| I125: Migrate: cleaning up migration state.

2018-02-01T00:40:48.495Z| vcpu-0| I125: SVMotion: Enter Phase 13

2018-02-01T00:40:48.495Z| vcpu-0| I125: SVMotion_Cleanup: Scheduling cleanup thread.

2018-02-01T00:40:48.495Z| Worker#1| I125: SVMotionCleanupThread: Waiting for SVMotion Bitmap thread to complete.

2018-02-01T00:40:48.495Z| vcpu-0| I125: Msg_Post: Error

2018-02-01T00:40:48.495Z| vcpu-0| I125: [vob.vmotion.stream.keepalive.read.fail] vMotion migration [a3e007a:1517445513843299] failed to read stream keepalive: Connection closed by remote host, possibly due to timeout

2018-02-01T00:40:48.495Z| vcpu-0| I125: [msg.checkpoint.precopyfailure] Migration to host <ip.add.re.ss> failed with error Connection closed by remote host, possibly due to timeout (0xbad003f).

2018-02-01T00:40:48.495Z| vcpu-0| I125: ----------------------------------------

2018-02-01T00:40:48.559Z| vcpu-0| I125: Vigor_MessageRevoke: message 'msg.checkpoint.precopyfailure' (seq 219224) is revoked

2018-02-01T00:40:48.559Z| Worker#1| I125: SVMotionCleanupThread: Waiting for SVMotion thread to complete.

2 questions:

What would be the underlying cause of the failure?

if no good reason for failure then how can i set the default migration to drop the VFRC contents?

[kabanossi]

First off, check if you have the sufficient amount of free space in target host's flash pool.

Double check your vFRC settings: VMware Knowledge Base

Also, don't forget that DRS will not migrate your vFRC-enabled VMs in most circumstances. DRS only moves such VMs in case of critical overprovisioning or "Maintenance mode" (despite it fails in your case).

[EddiePhoenix]

Kabanossi, Thanks for the insight.

I'm migrating to and from hosts with more than enough available Flash resources to handle the VMs migration if i didn't the migration would fail before it started. My settings are correct. And DRS will migrate a vm when the host is going into maintenance mode which i did mention was the case in this issue. However none of this is relevant to my questions. i probably should have included a little more info in the post. I will update it.

rephrase of the questions:

what would cause a vFRC migration to fail mid migration and require a restart of vCenter and the host to correct the crash?

how do i make the default vFRC migration drop the cache?

thanks again for your thoughts, got any more?!?!?