Strange gLiveView behavior after 1.26.1

Thanks for the update. I had TraceMemPool turned off on all my relays for many epochs with no problems. I have now turned it off for the BP’s as well. Fingers crossed that I have the same success as you are seeing. Our colleagues over at SmaugPool reported this type of behaviour previously: [BUG] - TraceMemPool set to true lead to missed slots on block producers · Issue #2565 · input-output-hk/cardano-node · GitHub I have missed a couple of blocks this epoch because of this unfortunately.

Same behaviour happens also if you set "minSeverity" other than Info in mainnet-config.json

I’ve noticed that if setting TraceMempool to false at the BP node, Processed TX and Mempool TX/Bytes are always zero. I believe we need Processed TX increasing in order to mint blocks at BP, so TraceMempool can’t be false for the BP or am I missing something?

Yes. It just displays as zero but the work is still being done correctly in the background. Completely safe.

Thanks @mcvetyty.

Is there a way to check if the BP is working fine (ready to mint) at the background level? Is there a specific systemd journal or another log entry I could search for?

No, not that I am aware of. I have all my servers set to false and am expecting a block this evening. I can report back that things worked as expected if you’d like but I have seen others post that they also have this set to false with no issues.

Sure, it would be nice to have a confirmation that everything is working fine. I’ve changed mine to false on both BP and relays but since I didn’t have a chance to mint a block yet, I really don’t know what to expect. Thanks again :slight_smile:

same here. I’ve seen it few times as well since 1.26.1 update

Confirmed - My BP minted a block last night with no issues and it had TraceMemPool set to false.

1 Like

I confirm as well. My pools have been consistently minting since changing the setting on the BP’s yesterday. Memory usage is also staying stable. Other than missing the visual cue of seeing TX’s increasing with gLiveView, the benefits of a more stable node is certainly a better choice.

But then - how to monitor if a BP node is working properly. I used to check diffs of TXs, but with TraceMemPool set to false that’s no longer possible.

Would diffs of slot numbers be ok?

Thanks @mcvetyty @MantisPool-MANT for your feedback.

I suppose that if you check out your logs folder at the BP server i.e. /opt/cardano/cnode/logs and issue a tail -f node0-****.json (the most recent json log file) you will see slot and chainDensity increasing, it could be an indicative that the BP is keeping up with the network. But I am not really sure that it also means that the BP is ready to mint new blocks. Suggestions anyone?