so basically, you can still do this step after you upgrade your version to 1.26.1
Hi Alex ,
Thank you for your fast update guide. I have a question regarding to prereqs.sh
I see that line → export BOOTSTRAP_HASKELL_GHC_VERSION=8.10.2
but i´m not sure if would be better update it with the version 8.10.4 of ghc. (I see that in other infos).
Do you know something about that?
cheers
Thanks Alex! works like a charm!
it’s up to you if u have to update the ghc version… I will stay with 8.10.2, I don’t think is a special request to have 8.10.4
You are right:
Ola Ahlman | AHL | Principal Maintainer of CNTools & Guild LV, [07.04.21 18:43]
Doesn’t really matter, 8.10.2 works fine
Im currently at 1hr 20mins on “starting”. Has your synced yet?
Yes, wait… needs to re-sync the DB… could take 2 hours… let me check how long it took for my relay … ~2 hours
Word, thank you as always
Thank you for the insanely quick post/update @Alexd1985 !
For AWS users:
You can create AMI snapshot of your nodes to back it up (snapshots may restart your server - some downtime)
I have not tested this method as this is my first upgrade. But I’m thinking of trying to following below so there will be no downtime:
- From AMI snapshot of a relay, deploy it an EC2 instance as a standby BP.
- Perform update on Relay node to cardano-node 1.26
- Do some final checks and test to make sure node is functioning properly
- Add in the BP files (vrf,kes,node cert) to standby BP but do not restart the node
- Update Relay/BP topology to include the standby BP
- Stop the old BP node
- Restart the cardano-node on new standby BP node
Let me know if anyone has had experience with updating and whether method above works or what you’ve done.
Hope it helps, thanks
David
Should work very well
Took about 55 minutes to sync and finally good to go! Patience everyone!
sudo cp $(find $HOME/git/cardano-node/dist-newstyle/build -type f -name “cardano-node”) /usr/local/bin/cardano-node
this doesnt work for me?
the previous command worked
Weird… my relay is running more than 4 hours but still stuck at “Status: starting…”
Service status tells me “Failed to query protocol-parameters”

Socket path is empty…
Any idea?
Try journalctl -e -f -u cnode.service
4 hours I think it’s to much but for me it took ~2hours with 8vcpu
Already checked, nothing special:

Looking at the linked Github Issues it seems like TraceMemPool behaviour should be better in 1.26.1. But its not 100% clear if there is still a risk to loose blocks. Is this trace needed for anything?
Here is a link which asks the same question bot with no answer so far: [Documentation] - TraceMempool: practical usage for relays · Issue #2350 · input-output-hk/cardano-node · GitHub
Thanks for the notes, @Alexd1985. Everything worked great for me. 3 relay and 1 block producer. Took about 2 hours.
For 1.25.1 - I did disable the Trace Mempool. I haven’t switched it back yet so I may still experiment with that.
1.26.1 has a significant CPU improvements, I can tell u for sure
But its not 100% clear if there is still a risk to loose blocks.
If ur node has not enough HW resources with 1.25.1 u can loss blocks because of high CPU usage when is set to true
Is this trace needed for anything?
Yes, for example u can track the TX processed
1.26.1 has a significant CPU improvement… u can set to true now
Nothing changed?