Nope, should not be any issues… do u have missed blocks?
Where do i have to check for missed blocks?
Where do u check for assign slots?
U don’t use cncli ? This script will show u if u have blocks assigned for the current and next epoch
I’ve just started using cncli with the new coincashew stake-snapshot approach to avoid RAM overloads in the past and the output is all 0 except 0.14 expected.
This all thing started today after someone told me he checked my pool and I had a scheduled block for current epoch, but I can’t find any confirmation to that…and checking slot leader schedule should be possible from BP only, by SPO only, for your SP only, right?
If that person has your pool files, then he can check
POOL_ID , POOL_VRF_SKEY and POOL_VRF_VKEY
No one has my pool files except me, 100% sure about that. I’ll simply take that sentence as a wish to me, a lie or a wrong check.
Back in topic, none of the relays showed any missed slots but BP showed that 1324 and increasing value.
I set Tracemempooltx to FALSE on BP, restarted BP and still no slots missed, more than 1Gb RAM freed and keeps looking good.
My BP is running on the oldest and reasonably slower Mobo-Cpu pair and slowest buses and RAM clocks of the 3 baremetal servers, could that have been slowing down the BP too much and making me miss slots or should I consider adding RAM to all the servers for the future?
I will expect that the nee version should use less RAM so… u can wait for the moment
I did the upgrade as well. Looking good so far. 22 hours without missing slots.
Missing slots during epoch changes is normal and documented. IOG is working on mitigating the issue. It’s a lot better compared to 1.26.1.
2 Likes
Actually, you can compute leaderlog from any computer. You only need a synced node (like with Daedalus) and a synced cncli database, the vrf key, and the pool id.
16 missed slots in 24 hours, sensibly lower but still present…
What is the RAM configuration? … I saw now u have 8G + 4G swap but how is the CPU?
also, u can try to start the cloud server as a Producer and monitor it for few days (and the actual
BP start it like a relay)
CPU is an AMD athlon x2 2C/2T at 3.2 GHz.
The main question to balance my CPUs through my nodes is: which node has to be the most powerful?
While setting up everything I read through different opinions on it, who said BP due to minting, who said relays due to connections management…
IMO bp node requires better CPU, because of some “real time” requirements of processing all transactions when being the slot leader. relay requires more memory because it has to handle more connections.
However, 16 GB for relay is enough for now. And a bare metal bp node should be enough as well.
1 Like
In my opinion, your CPU is a little bit weak in some moments of pool operation.
If possible, try upgrade CPU for at least 4c/8t and RAM for 16 GB, so it can handle the load spikes better.
1 Like
Since I have one 2C/4T and one 4C/8T relays, I’m considering upgrading 4C as BP and downgrading 2C/2T as a relay, and I’m confident I’ll be able to do that by simply switching SSDs between the 2 and Ubuntu will do the rest…hopefully!
Otherwise, I could buy a compatible 4C/8T CPU for 30usd and send the 11years old 2C/2T CPU to dumpster, if it’s not suitable even to be a Cardano node 
Thinking back, I also had some issues with that CPU when testing before deployment… installing Ubuntu desktop caused some weird video issues since installation wizard, never seen anything like that in 20 years and server was unusable with GUI.
Ubuntu server was installed and had been running fine since the beginning…or this is what I was thinking till 2 days ago!!! 
Looks like Missed slots keep on showing up, so i’m figuring out different ways to make the relay work as a producer as quickly as possible.
In order to start a relay as a producer, should I simply change firewall settings, topology, remove topologyupdater and cronjobs add the pool files to that relay before restarting the node with a producer script, correct?
What if I switch SSDs one another? Do you guys think Ubuntu is going to work fine?
the Ubuntu might work, but be prepared for some network setting adjustments
Everything worked fine switching SSDs between servers, just had to set new ethernet config and worked straight away, thanks!
Anyway, I migrated BP to 4C/8T 2.93Ghz, sensibly updating computing power, but still missing around 30 slots in 24hours. What can I investigate? RAM?
Another hypothesis: BP and 2 private relays are connected to same unmanaged switch which could (should) prioritize performances based on traffic load. So the BP is missing slots only from cloud relay due to that switch.
Hi, I’ve resolved starting the BP with this parameters +RTS -N -RTS that says to the cardano-node to use all cores availlable in the machine. The default is 2 I think. You can say to use for example only 4 cores with -N4. As a side effect the memory usage is increased, but no more missed slot, only some on epoch change.
I solved it with a combination of Hardware Upgrade, Garbage Collection settings, and the “-N” flag. See link below for what I did. Maybe this will help some others in a similar situation.