overlay survey should look and act more deterministic #2592
Assignees
Labels
Comments
bboston7
added a commit
to bboston7/stellar-core
that referenced
this issue
Jan 31, 2024
This change adds a `simulate` command to the network survey script that takes a network topology in graphml format (such as the one produced by the `survey` command) and simulates a survey of said network. It is intended as a tool to use during development of the script itself to test the script without running a survey on the network, and should make it easier to address issues such as stellar#2592. This is an initial version of the `simulate` command that will evolve as we need to test the script under additional scenarios.
6 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Right now when running a survey, even with a small timeout (like
-d 30) the script can run for potentially a long time.We should log some stats to show that it's doing something (like backlog size, number of peers and connections discovered).
Also, the criteria for stopping the script is odd: it uses core's internal survey state to decide on this.
In particular, this seems quite arbitrary as the script may retry many times to survey certain nodes.
I think it would make more sense to have the script actually decide on its timeout based on its own backlog, and use core's internal timeout more as a mechanism to make sure that even if the script crashes, core stops surveying.
The text was updated successfully, but these errors were encountered: