There seems to be still a lot of issues and / or unclarities stemming from the agent. Some can be solved with bugfixes, some can be solved with documentation. It's alpha / beta software anyway. But we're moving in testnet phases where this software, in particular the reward lifecyle management, will be quasi critical.
As far as subgraph management goes, i not only seem to be running much more reliable / less issue prone without, i'm also a lot more flexible / customisable without. Consider in particular heterogenous indexerpools : maybe one is not as performant, more expensive to run, etc but still make for useful failovers or spare capacity.
Thus i'm hoping that the subgraph management can be made optional, so we can use it for reward lifecycle management solely. And no doubt the agent too, including subgraph management, will over time become more proficient / clear / stable. Though even then, people might prefer to run without or roll their own solution.
There seems to be still a lot of issues and / or unclarities stemming from the agent. Some can be solved with bugfixes, some can be solved with documentation. It's alpha / beta software anyway. But we're moving in testnet phases where this software, in particular the reward lifecyle management, will be quasi critical.
As far as subgraph management goes, i not only seem to be running much more reliable / less issue prone without, i'm also a lot more flexible / customisable without. Consider in particular heterogenous indexerpools : maybe one is not as performant, more expensive to run, etc but still make for useful failovers or spare capacity.
Thus i'm hoping that the subgraph management can be made optional, so we can use it for reward lifecycle management solely. And no doubt the agent too, including subgraph management, will over time become more proficient / clear / stable. Though even then, people might prefer to run without or roll their own solution.
Kindly