Profile a log file
Give Logcaliper a sample of the log you plan to collect. It reads the file in your browser, line by line, and reports what it is made of: how many lines and how long they run, the message templates behind them, the events per second over the time the sample covers, and what a year of it takes to keep. Nothing is uploaded.
Logcaliper cannot read this file
The reader stopped
Files of 50 to 500 MB are the normal case.
No log at hand? Try the sample: 1.5 MB of made-up syslog, fetched from this site and then read like any file.
In progress
Reading
Result
Profile
What this log takes to keep
The same arithmetic over a list of sources, with replication and the number of nodes, is the List and Storage tabs of Logcaliper, the free iOS calculator; the measured event sizes it ships with come from samples like this one.
| Lines | Template | Share | Avg bytes |
|---|
How this works. The file is read in chunks of about 8 MB by a Web Worker, so the page stays responsive while it runs. Every line's byte length is counted, terminator included, so the totals reconcile to the byte against the size of the file. Timestamps, addresses, ids and numbers are masked, then Drain, the log parsing algorithm, groups each line with the template it shares at least 40% of its words with and wildcards the words that differ; up to 4,000 templates are kept, and past that the least recently used are folded into Other. The rate is the lines that carry a timestamp over the span from the earliest to the latest, in whichever of six common formats the first 200 lines use. The storage line is the arithmetic of the app's Storage tab, at the retention and compression you pick. In this browser the worker was loaded from worker.js directly rather than built inside the page; it is the same code, and after a stop that had to shut it down it is loaded again the same way.