[Modbus] Poller crashes after few hours ModbusIOException

Expected Behavior

Poller keeps polling and data should be provided at all times.

Current Behavior

I have a DAT11 configured as Modbus slave on TCP port 1800. In OH I have configured the Modus TCP, Poller and Data. After a few hours of getting data the poller crashes with this error:

Error with read: org.openhab.core.io.transport.modbus.internal.ModbusSlaveIOExceptionImpl: Modbus IO Error with cause=ModbusIOException, EOF=false, message='I/O exception - failed to write: Broken pipe', cause2=null

And this in the openhab.log:

2024-09-03 12:29:11.577 [ERROR] [rt.modbus.internal.ModbusManagerImpl] - Last try 1 failed when executing request (ModbusReadRequestBlueprint [slaveId=255, functionCode=READ_MULTIPLE_REGISTERS, start=0, length=9, maxTries=1]). Aborting. Error was I/O error, so resetting the connection. Error details: net.wimpi.modbus.ModbusIOException I/O exception - failed to write: Broken pipe [operation ID f5988cac-492c-4462-8dcf-db5d76b869f0]

When OH is generating these errors, I can read out the register on the DAT11 just fine with QModMaster. So this must be bug related in OH.

Possible Solution

Don't know.

Steps to Reproduce (for Bugs)

This is how I configured my things:

Slave

UID: modbus:tcp:cea7a1d77b
label: PavoneSystems - DAT11 - Y - Slave
thingTypeUID: modbus:tcp
configuration:
  rtuEncoded: false
  connectMaxTries: 1
  reconnectAfterMillis: 5000
  timeBetweenTransactionsMillis: 1000
  port: 1800
  timeBetweenReconnectMillis: 5000
  connectTimeoutMillis: 30000
  host: 192.168.111.22
  afterConnectionDelayMillis: 5000
  id: 255
  enableDiscovery: false

Poller

UID: modbus:poller:cea7a1d77b:850c5d06fc
label: PavoneSystems - DAT11 - Y - Poll
thingTypeUID: modbus:poller
configuration:
  length: 9
  start: 0
  refresh: 10000
  maxTries: 1
  cacheMillis: 0
  type: holding
bridgeUID: modbus:tcp:cea7a1d77b

Data

UID: modbus:data:27284b8263
label: PavoneSystems - DAT11 - Y - Gross weight (MSB)
thingTypeUID: modbus:data
configuration:
  readValueType: uint16
  readTransform: default
  writeTransform: default
  writeType: holding
  readStart: "2"
  updateUnchangedValuesEveryMillis: 1000
  writeValueType: int16
  writeMultipleEvenWithSingleRegisterOrCoil: false
  writeMaxTries: 3
  writeStart: "100"
bridgeUID: modbus:poller:cea7a1d77b:850c5d06fc
channels:
  - id: number
    channelTypeUID: modbus:number-type
    label: Value as Number
    description: Number item channel
    configuration: {}

Findings

During the OH errors we can poll and retrieve data from the DAT11 with other software.
When I pause & start the Modbus TCP or Poller thing in OH WebUI, the error remains.
If I reboot OH, the issue is resolved.
When we reboot the DAT11 device, the error is resolved. Perhaps this is because something resets in OH and therefore it works again. As the DAT11 just keeps working during this errors.
Had these errors from the start when using the DAT11, starting on OH3+

Your Environment

Docker image on Intel NUC (had same errors when running as application in OS)
Ubuntu 22.04.2 LTS
OH 4.1.3 release build

I created a topic in the community here: https://community.openhab.org/t/modbus-poller-crashes-after-few-days/158299

openhab / openhab-addons