net-p2p/ogmios: Review rc script

[arrowd]

• Do we need to create a user for the daemon?
• Why it uses a real cardano-node socket file under jail instead of a public symlink?
• Is it OK to have -o ${logfile} in the daemon call?

[reqlez]

Reason uses real socket is because i am mounting the whole socket directory via nullfs, this is why i changed cardano-node to have socket under /socket/ directory instead of jail root.

I think one of the commits has some information about this.

ogmios does not have any other logging configuration as far as i'm aware? But i can look into this. I know cardano-node and cardano-db-sync i configured logging via json file.

[arrowd]

Reason uses real socket is because i am mounting the whole socket directory via nullfs, this is why i changed cardano-node to have socket under /socket/ directory instead of jail root.

Ah, right. And the ogmios itself runs under a jail, so it needs a nullfs mounted into it.

ogmios does not have any other logging configuration as far as i'm aware? But i can look into this.

The problem with -o ${logfile} is that these logs aren't rotated and can grow up to enormous sizes. If there is nothing useful in these logs, it can be removed. On the other hand, if logs are useful, but scarce, they can just be redirected to stdout.

[reqlez]

The problem with -o ${logfile} is that these logs aren't rotated and can grow up to enormous sizes. If there is nothing useful in these logs, it can be removed. On the other hand, if logs are useful, but scarce, they can just be redirected to stdout.

Let me review what kind of output there is, and logging options and i will get back to you.

[arrowd]

Any progress on this?

[reqlez]

Any progress on this? oh yea... i completely forgot about this one... checking now

[reqlez]

So the logging is useful... however, there is not setting that i can do in config file or anything to redirect logs to something else.

How do you want to handle that ? Logs are not extremely spammy or anything it seems.

[arrowd]

How about cleaning them on the service start so that they wouldn't pile up?

[reqlez]

How about cleaning them on the service start so that they wouldn't pile up?

not a bad idea for now... unless we figure out a rotation later. but clear before the service actually starts since we need start-up logs if things don't work.

I honestly think we should do the same for cardano-node and db-sync ... if config file is bad, it will output to console and people won't be able to figure out why cardano-node is not starting.