You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
it has been observed that, after starting the devnet and sending transactions to it, there is a point of time (which depends upon the amount of transactions that are sent) where the chain starts to slow down.
This is observed by:
Increase in block time
Decrease in TPS
Increase in disk reads
Several metrics have been collected, with Grafana, that show this problem.
We are currently investigating if this is a problem with our code or the data structures that are used to store information in the database, or the database itself.
It is important to mention that this screenshot was taken from the minimum scenario where this has been observed, which is a client pushing transactions to itself which rules out the possibility of a network problem, and also, this is why such a huge amount of TPS is observed at the beginning, which does not correspond to a more realistic scenario where there are multiple nodes over a network and some other nodes sending transactions to the network (although the issue has been observed in this latter scenario as well)
The text was updated successfully, but these errors were encountered:
The following screenshot, shows the high increase in disk reads when the issue starts to occur
Also it is easy to note the correlation from high increase in disk reads and the blockchain slow down.
Marking the issue as solved, as the performance degradation that is now observed, after introducing the NORDAHEAD flag is just a consequence of running out of memory (disk cache) and disk space, which can occur if a full history node does not have enough resources to maintain all the data that is generated.
it has been observed that, after starting the devnet and sending transactions to it, there is a point of time (which depends upon the amount of transactions that are sent) where the chain starts to slow down.
This is observed by:
Several metrics have been collected, with Grafana, that show this problem.
We are currently investigating if this is a problem with our code or the data structures that are used to store information in the database, or the database itself.
It is important to mention that this screenshot was taken from the minimum scenario where this has been observed, which is a client pushing transactions to itself which rules out the possibility of a network problem, and also, this is why such a huge amount of TPS is observed at the beginning, which does not correspond to a more realistic scenario where there are multiple nodes over a network and some other nodes sending transactions to the network (although the issue has been observed in this latter scenario as well)
The text was updated successfully, but these errors were encountered: