Avantis PLM+ Fail

I had 2 PLM+ fail during an event.
At the FoH I had an Avantis on Internal Clock 96 KHz with a Dantecard 64 x 64. The connection to the stage was 2 x Cisco CBS-350 connected via a trunk. The switches have the usual QOS settings and IGMP snooping was turned off.
At the stage there were 3 x PLM+ 12K44 2 x Shure AD4Q 2 x DT168 connected via Dante.
At the same time, I had Shure control and the Avantis control network running with manual IPs in the same VLan (Networkbridge on the Dante card was off)
During the show the system started cracking, by which point it had been running for over 30 hours. Then first PLM said goodbye and after about 5 minutes the next one.
Restarting the Avantis made everything work again. I set up the same setup in the warehouse and it played for a week without any problems.
Here is my log from Avantis. Can someone help me with this? is sync to external the solution?

Sat 23.11.24 21:27:36 TouchScreen - Update Version String UF V2.0
Sat 23.11.24 21:27:08 TouchScreen - Update Version String Version Unknown
Sat 23.11.24 21:26:55 TouchScreen - Switching to Internal Clock
Sat 23.11.24 21:26:54 TouchScreen - TS InitialEnquiry Complete 0
Sat 23.11.24 21:26:54 TouchScreen - Main Board ID - 32
Sat 23.11.24 21:26:54 TouchScreen - Input Audio Board ID - 2
Sat 23.11.24 21:26:54 TouchScreen- System Started - V1.32 - Rev. 84450
Sat 23.11.24 21:26:54 TouchScreen - Connection made to 127.0.0.1 (AHNet)
Sat 23.11.24 21:26:54 TouchScreen - Outgoing Remote Connection: 1 - “127.0.0.1”
Sat 23.11.24 21:26:54 TouchScreen - Application Started - V1.32 - Rev. 84450
Sat 23.11.24 21:26:53 TouchScreen - cApplication created
Fri 22.11.24 10:39:54 TouchScreen - cShowManagerClient::SendRecallTask: M_RECALL_SHOW Factory: Show:"
Fri 22.11.24 10:37:42 TouchScreen - cShowManagerClient::SendRecallTask: M_RECALL_SHOW Factory: Show:"
Fri 22.11.24 10:36:01 TouchScreen - Update Version String UF V2.0

Hard to say but it sounds like either latency increasing at the receiving end (PLM) due to network issues, or clock drift to the point of being out of tolerance - either of these could conceivably lead to a mute in the PLM. Dante Controller offers diagnostic tools you can use if this happens again (latency monitor, clock screen and Dante Controller logs would be my go-to places for troubleshooting before looking at Avantis).
What was the clock setup for the Dante devices and was the Avantis set to Internal clock or to sync from Dante?

Unfortunately I didn’t have the Dante controller running, so I don’t have a log. I set up the system again and checked the latency, everything was fine. I had not choose a preferred leader in the Dante network and had not set “sync to external” on the Avantis and the Avantis was set on Internal Clock. I suspect that was the error. The PLMs showed me clock slipping when they were muted.