I have searched but couldn’t find any specific relatable threads. So I hope someone can point me in the right direction.
We started the upgrade for Vasil last friday (Hurrayy). Everything went okay, until the synchronisation. It is still running since friday on 16GB ram hosting cluster. When I look at system load, I see CPU constant 35 -45 % usage, and RAM also about 40 - 60% with a sudden temporary increase to 80% today.
These are not numbers that I would expect if hardware or internet speeds was the problem.
How come it is soo terribly slow to sync, and what can be done to increase it ? Should I add more CPU cores, but not sure how this would help if it’s not heavily loaded.
Anyone here who has the same experience?
It is not syncing, it is rebuilding the ledger database (re-scanning all the blockchain from the db/immutable files). Internet speed does not matter, only CPU and Disk speed is important.
You should watch the logs of the service to see what exactly the service is doing. Usually, on a good server / VPS with SSD or NVMe disks, it should take between 2 and 4 hours, if the server / VPS does not have a good configuration (the disk probably affects the most), it can take a lot more. But it should not take days. Check the logs.
We have checked the log, but we see no errors, or anything out of ordinary. Its processing blocks, and that all seems to go well. Just incredibly slow.
Should we look for anything specific ?
I had the same issue a few days ago with setting up my new pool. It was at 92% and was already 2 days. What I did is just copied my local db from home to my vps which only took about an hour to complete.
I would search for an indicator of the progress (percentage).
What type of disks has the VPS?
When I upgraded my nodes, I upgraded one (which is a hardware server with NVMe disks), the upgrade took about 2-3 hours, and then I synced (rsync) the DB folder from it to all the other nodes, to speed up A LOT the process of upgrading.
For us the solution was uploading csnapshots. This saved us literally days of syncing. The logs where fine, mentioning progress, CPU and Memory usage all within acceptable limits. Nothing out of ordinary. But with this solution it was done within an hour. Thanks for pointing that service out.