Use a number separator for parsec shard##_endpoint to prevent double-digit confusion

[rockett-m]

Affected Branch

trunk

Basic Diagnostics

• I've pulled the latest changes on the affected branch and the issue is still present.

• The issue is reproducible in docker

Description

For endpoints in parsec, the option parsing for the following machines {agent, ticket machine, and shard} goes like this:

src/parsec/util.cpp, line 63

for(size_t i = 0; i < count; i++) {
    auto ep_key = component_name + std::to_string(i) + "_endpoint";
}

• Agents aren't replicated, so agent0_endpoint to agent{n-1}_endpoint works.

• The ticket machine can be replicated but there is typically just one, so a single number works.

• We can have many shards, and shard replicas; however, so the above code should be modified for double-digit compatibility.

In the trivial case (1 shard, 0 replicas) the representation shard00_endpoint is fair enough.
For many shards; however, we should use an underscore to separate the values like 2pc does.

opencbdc-tx/2pc-compose.cfg

Line 18 in 96062b5

shard0_0_endpoint="shard0:6666"

For example, we can't tell which shard or replica shard111_endpoint refers to if we have 50 shards and 15x replication. The underscore fixes that: shard11_1_endpoint. Even if we don't plan on double-digit replication, the separation helps readability.

The potential for many shards can be seen at the bottom of the diagram.

Code of Conduct

• I agree to follow this project's Code of Conduct

[rockett-m]

@HalosGhost I have the code for a proposed PR. Could you assign this to me? Thx