mazurel / analyzer Goto Github PK

Essentially, we do not want timestamps when comparing log files

Summary

We need a tool that, based on some config, generates two simple log files (per one run) which we can use for testing purposes.
There should be a small difference (for example: few lines with a new template) between these logs (based on configuration)

Configuration options

For example, we could have the following options:

• Maximal line length (in words)
• Minimal line length (in words)
• Template wariations (how many templates there are). Each template consist of some sequence of words + numbers, etc
• Number of lines
• etc

There needs to be some kind of tabbed menu for configuring heuristics (and drain in the future, see #4).

Each heuristic should be able to specify what kind of settings does it need and query their state.

It can be based on Nicegui tabs, but any solution is acceptable.

It should be visible for user, based on which log lines from another file time heuristic was calculated. This change should:

• Add an ability to share datastructures from heuristics
• Implement log "diff" view
• Connect all of this together

Implement heuristic, that based on clastered logs (templates) would use diff-like algorithm for comparing log line importance

EDIT:
This should contain also (mainly) changes to view. We want to see log files line by line with customized diff

EDIT2:

• Initial implementation: #36
• Improve initial implementation: #39

• Line numbers
• Better colors
• Logs filtering, etc

(cache position of timestamp inside given log file)

Please ensure that unit tests work and update them accordingly

I think this issue should be implemented after #3

• Asynchronize UI (#32 )
• Asynchronize heuristics (#34 )

We need to collect logs from easy travel and use them to verify how well our solution works.

We need to run application N times for T time, where T should be at least 6 hours. Also T needs to be the same for each log case.
N is the number of cases that we want to cover, for example:

• Case when application works normally
• Case when there are many users
• Case when something crashes
• etc

Please label collected logs appropriately.

For example, the user may choose to use:

• Drain
• Brain
• Something else

With the same heuristics (they should be template miner agnostic)

This will require using: https://github.com/logpai/logparser