I've been investigating some unusually large log files in UMH-core (150+ lines repeated every 100ms) and wanted to share what I found. The Problem: During reconciliation cycles, the system is repeatedly logging the entire protocol converter configuration at INFO level, even when nothing has actually changed. This results in massive log bloat. What's Happening: Every reconciliation tick (~100ms), you see repeated log entries like this:

[INFO]  [service/manager.go:641]  [ManagerServiceBridge]  
Connection config: [{FSMInstanceConfig:{Name:service-bridge DesiredFSMState:up} 
ConnectionServiceConfig:{NmapServiceConfig:{Target:10.0.0.1 Port:1883}}}]

[INFO]  [service/manager.go:642]  [ManagerServiceBridge]  
Dataflow component config: [{FSMInstanceConfig:{Name:read-service-bridge DesiredFSMState:active} 
DataFlowComponentServiceConfig:{BenthosConfig:{Input:map[protocol:map[...]] 
Pipeline:map[processors:[...   150 lines of actual  config content    ...]]}}]

This exact same config block repeats identically multiple times per second. Root Cause: The logging statements dump the entire configuration structure at INFO level unconditionally, regardless of whether any actual changes occurred. The configs are identical across all the repeated log entries—nothing is actually being updated. This causes unnecessary performance impact Any idea if this is related to our specific setup? I suppose this should only be logged when actual config changes have happened? And only log summary (statistics) on INFO level?

https://www.umh.app/insights/changelog/v0-44-29 have you tried the newest version?

hi @Jermuk , problem is solved indeed. however i do still get a lot of repeated messages from processes that aren't ready yet, but in reality a write-DFC is not even configured for instance. To give you some context, we are trying to run the system on minimal hardware, examining which services are taking up the majority of resources, trying to disable what we don't need etc...

Hey @pascaldesmet. , we also saw these "false diffs for write-DFCs" and have already merged a fix for that. It will be included in the next release, which will come today or tomorrow. On top of that, I have been working on making the logs less noisy in general. This work is currently in review, so not sure if it will land in this next release or one after that. What this will include: 1. Fixed width columns for logs, so they are easier to read 2. If large amounts of the same log lines appear, they will be downsampled so that they don't flood the logs 3. The SnapshotLogger will only report a new snapshot, if there is an actual change. If the Snapshot stays the same, then it won't be logged 4. Lots of unnecessary

Info

levels have been downgraded to

Debug

, so less clutter Hope this will help you. I will ping you, once the new releases have landed! Thank you very much for your feedback!

@yannic_umh i have been offline for a few weeks. Planning to evaluate this soon. Thanks!