Really long log lines

Issues Dgraph
1

Experience Report for Feature Request

When using a cluster that does alot I am faced with reeeeally long log lines for schema and some errors. Some of the key offenders of single line messages are:

  • 138017 characters long: state from zero
    • This shows the entirety of the schema with thousands of predicates, gets way too much.
  • 408526 characters long: error applying proposal
    • This shows my whole big chonker mutation batches, at 400k characters thats… too much

What you wanted to do

Read the logs on a big cluster

What you actually did

Use bash-foo to truncate long log lines

Why that wasn’t great, with examples

Being able to use the raw logs would be nicer

Any external references to support your case

n/a

1 Like
2

Would be nice to have a flag to control these, @hardik

2 Likes
3

I would love to see this changed in the future as I currently have 2,674,221 characters in the membership state log line. 2.6 million characters in a single line is too much.

I can implement it if it will be taken. Maybe print out aggregated stats about the membership instead of the whole thing?

1 Like