FAQ

Answers to frequently asked questions (FAQs) about Lily.

Setup #

Do I need Lotus to run Lily? #

No. Lily does not need a separate Lotus node running since Lily itself is an abstraction over Lotus.

However, Lily can use a datastore created by Lotus.

Do I need to use TimescaleDB with Lily? #

A database instance is not required to use Lily. You can simply have a Lily node running and then run a walk job over it with CSVs as a Storage mechanism. These CSVs can then be loaded onto the database of your choosing.

How much storage should I allocate for my database? #

Around 4TB should be sufficient for mainnet data with a data retention policy of 3 months.

How do I configure Lily? #

You can configure Lily with the config.toml file.

Where do I get the config.toml file? #

Lily creates a config.toml file for you upon lily init if no configuration file is provided to the --config option.

Lotus also provides a default config.toml file with the handy lotus config default command. You can then redirect this into a file.

Why didn’t my config changes take effect? #

You’ll need to restart the Lily daemon.

Jobs #

How should I distribute tasks across jobs? #

Tasks should be grouped based on the actors they process for optimal cache utilization.

How can I backfill incomplete data? #

You can run the find job and then an adhoc fill job to backfill missing tasks at some specified range of epochs. This can also be run as a cron job.

How can I tell if Lily correctly exported to the database? #

You can use gap find to search for gaps in the database. You can also run the gap find function independent of Lily with SQL.

Why are my tasks failing? #

You can run gap find over the visor_processing_reports table.

All jobs that run tasks, [walk, watch, fill], will write the results of the tasks to visor_processing_reports.

Why is my job failing? #

Check lily job list to see the reason for job failure.

Can a job be successful even when tasks fail? #

Yes. A job runs a set of tasks and tasks can fail while the job is successful.

Syncing #

Do I need to let Lily sync until the current epoch? #

Yes. If you run jobs before Lily completely syncs, there will be unexpected behavior.

How can I make sure that Lily finishes syncing? #

lily sync wait && lily ...

What’s a good way to check that my repo has all of the heights I need for a walk? #

lily job run --tasks=actor_states,consensus,messages walk --from <min_height> --to <max_height>

Why do tasks in my job fail? #

This can be for multiple reasons. But the most common ones are:

• your Lily node is not fully synced
• the snapshot you used to initialize Lily does not have the state of either your minimum or maximum epoch height