-
Notifications
You must be signed in to change notification settings - Fork 1.7k
pruning archive node cannot breach more than 2.0blk/s super slow 2.7.2 #11502
Comments
@ApolloMuses we are seeing several reports from other users on tracing and/or archive nodes with similar issues. We are still investigating the issue and collecting information from reporters, see e.g. #11494 or #11503. Is your node also under RPC traffic while syncing? Can you try running the node without specifying the cache size (i.e. remove |
Thanks for the quick response. Here's the log below: Note the path was a bit different, I didn't have the
I'll try the remove cache again to see if it makes a difference, I removed it previously and it didn't really affect syncing performance. In terms of usage: CPU: ~3%. Quite low. RAM:
|
The node has minimal to no RPC traffic. So I can't imagine that being the same issue. After removing the cache, the parity process hangs. Now I can't even get the process to start with or without the cache tag...
|
@dvdplm David, any ideas on how this can be resolved? |
Configuration script used:
parity --allow-ips=public --max-peers=128 --max-pending-peers=128 --cache-size=32768 --no-serve-light --no-periodic-snapshot --no-secretstore --scale-verifiers --jsonrpc-apis="web3,eth,pubsub,net,parity,parity_pubsub,traces,rpc,secretstore,shh,shh_pubsub" --jsonrpc-hosts="all" --pruning archive
Currently synced to ~7,945,500 Blocks and going at 1-2blk/s on average.
Hardware:
Ec2: M5.8xlarge w/ a 3.5Tb attached EBS volume.
This is equivalent of:
32vCPU
128Gb of memory
Bandwidth of 10Gps
Prior to upgrading to v 2.7.2, the node was crashing due to memory leaks, but upgrading seem to have fixed the issue, having said that the sync sleep is still agonizingly slow.
Anyone else experiencing this or have any idea how to speed up the sync time while still maintaining an archival node?
Before anyone ask, I've also tried the configurations on c & r typed instances and noticed marginal if not equivalent performance.
Any comments or suggestions would be much appreciated. Thanks in advance!
*the node was previously at one point fully synced but that was with a m5.24xlarge instance, but later on the node was downgraded to a less powerful instance which couldn't keep up with the network.
The text was updated successfully, but these errors were encountered: