Where are the "server logs" stored?

[jsaern]

I need to review the server logs to debug a crash during reindexing for 0.15.0.

I am unable to finish re-indexing for 0.15.0 as Hister crashes and advises me to read the server logs. I am unable to find the server logs in the configured directories or using fzf to find hister folders on /var.

Error from reindex sometimes happens within minutes, other times takes 10,000's of entries before failing:

Error! Reindex error: server error (500): Internal Server Error
Check the server logs for details

[asciimoo]

Currently there is no logfile. Logs go only to the STDERR of the listen command. However it would be good to have a configurable logfile. I'm implementing it soon.

Thanks for the question/suggestion.

[asciimoo]

I've added log file configuration option with 2d25f7b.

Please post the results of your findings, this reindex bug is really concerning.

[jsaern]

Thank you!
Log settings worked. I set it up to put it in a sub-folder in the hister data folder for now.

~/.local/share/hister/log/

• Log with default loglevel: hister.log

• Log with debug loglevel: hister.log

[jsaern]

The reindex operation fails pretty quickly, so I am running it one more time on the same running instance. Once it fails, I'll add that version of the debug log.

Watching the log, I wonder if you have already set a way to filter out all the actions that show up from using web applications such as Google Workspace or Umap.openstreetmap.fr.

[asciimoo]

Thanks for the logs.
Seems like an extractor abort is responsible for the premature reindex stop:

{"level":"error","error":"extractor Notion: notion page content not rendered","time":"2026-06-11T08:31:18-07:00","caller":"server/server.go:1767","message":"reindex failed"}

I've fixed it in 40267f9. Could you test it again?

[jsaern]

The latest commit reindexes much faster and has yet to stop.

Seems ok.

[jsaern]

The logs are less clear for reading directly than the STDERR output, though.

I liked how the STDERR output showed how far along you were (x/total).

[asciimoo]

Yeah the file based logs are in JSON format. It's easier to parse it with software (jq is great for this), but harder to read. I'm not sure which one should we prioritize. Perhaps the best would be a log_format config option to support both.

[jsaern]

So, for now, if I make this interim version my local default until the next release, I could "remove" the log_file setting to monitor the messages via STDERR. Or, I could learn how to use jq.

I'll wait until hister finishes reindexing and take a look.

Thank you!