Track down errors
When a log is full of errors, two questions come first: when did they start, and which error is the most common? LogLens answers the first with a per-minute chart and the second with error clusters. You can then save the lines you found to a file.
See when errors happen
Press t to show the error histogram, and t again to hide it.
The chart has 15 bars, one per minute, covering the last 15 minutes. Each bar is the number of error lines that arrived in that minute; the newest minute is on the right. The label reads errors/min · 15m · peak 4, where the last number is the highest bar. It updates once a second.
The histogram counts lines by when LogLens read them, not by the time written inside the line. The 50 lines from each file loaded at start-up all count as arriving at that moment, so the first bar can show a spike that did not really happen then.
Group similar errors
Press c to open the error clusters panel.
The title line reads error clusters — 40 errors, 6 clusters: how many error records LogLens found and how many groups they fall into. Each row then shows:
| Part | Example | Meaning |
|---|---|---|
| Count | [×14] | How many errors are in this group. |
| Title | ConnectionError: connection refused | A cleaned-up version of the error. |
| Sources | 2 sources | How many of your files it appeared in. |
| Last seen | 14:31:07 | When the newest one arrived. |
The group with the most errors is at the top. If the buffer holds no errors, the panel says no errors in buffer.
How errors are grouped
LogLens treats an ERROR line plus its traceback lines as one error record. Before comparing records it replaces values that change every time, such as timestamps, IDs, UUIDs, hex addresses, numbers and quoted text, with placeholders like <n>, <uuid> and <str>. That is why titles can contain those placeholders.
- Errors with a stack trace are grouped by the files and functions in the trace (line numbers ignored), the exception type and the cleaned-up message.
- Errors without a stack trace are grouped by their cleaned-up message. Messages that are almost identical are merged.
The panel looks at the whole buffer and ignores your filter and errors-only setting. It refreshes every two seconds while new lines arrive.
Jump to an error
Use ↑ and ↓ to pick a row, then press Enter. The panel closes, LogLens pauses, and the view jumps to the newest error in that group. Press p to go back to the live tail.
Close the panel without jumping with c or Esc.
If you see that cluster's newest line has been evicted, that line was pushed out of the 100,000-line buffer between the panel updating and your selection.
Export lines to a file
Export saves the lines that your current view lets through: the whole buffer, narrowed by the active filter and errors-only setting. It is not limited to what fits on screen.
Set up the view you want to save, for example press e for errors only, or / to filter.
Press w. The export box opens with a name such as
loglens-export-20261109-142530.log(the date and time now).Keep the name or type another path. A name without a folder is saved in the folder you started LogLens from.
Press Enter. A message such as
wrote 312 lines to loglens-export-20261109-142530.logconfirms it.
The file holds the original log lines, one per line, without LogLens's labels and times.
LogLens never overwrites a file. If the name is taken you see file exists: followed by the path, and the box stays open so you can change it. Any other problem, such as a folder that does not exist, is shown the same way. Press Esc to cancel.