Page MenuHomePhabricator

Create tools to retrieve relevant information from logs in a useful format
Closed, DeclinedPublic4 Estimated Story Points

Description

We often need to find information from one log line based on criteria matched on another log line within the same session.

The tools might take as input:

  1. search pattern to select sessions
  2. output line pattern, to select the line with the info we want from within the selected sessions
  3. output selection
    • delimiter and field numbers
    • regex patterns
    • xpath selections

Or we could bake in all sorts of recipies for selecting specific pieces of data from specific gateway logs.

Search parameters should also include start and end dates

Should have option to de-duplicate output by email address.

@awight, did we have an Etherpad going for this brainstorm once upon a time?

Use case:

We would like to use our email program to ask donors with failed payments to retry. This was tested with high success after the payment outage on 9/2/15 - the email had a 20% donation per email rate which is basically unheard of...

So we want to try this with more "normal" GC 100 failures following a Wednesday English banner test. The trouble is (per T111448) I don't know how to get the data. For the 9/2 failure, GC sent us all the order IDs, but I want to find a scalable process so that's not an option this time. Can someone teach me how to use grep to get email addresses, and ideally a few other things, for failed payments?

This is not urgent, but I would very much like to run this pilot by the end of October lest it is something I want to do regularly during Big English. Hopefully a short training session will be enough for me to manage the process alone going forward.

Event Timeline

CCogdill_WMF raised the priority of this task from to Low.
CCogdill_WMF updated the task description. (Show Details)
CCogdill_WMF subscribed.

There's some abandoned work here, my thought is that we build a tool to pull out log records matching a pattern.
https://gerrit.wikimedia.org/r/#/c/234417/

Hi @CCogdill_WMF! While learning grep would be fun, we may have to take a different approach here. Because those logs are on production, I guess we're trying to minimize the number of people with access to them. @awight is working on a tool that will make it very easy for people with access to fulfill requests like this in the future.

Cool, @Ejegg and @awight, that sounds great too! It also sounds like a huge project. Keep me posted if Sprint +1 actually feels realistic.

atgo subscribed.

@Ejegg @awight can you update the name of this task to reflect the new reality please?

Ejegg renamed this task from Teach ccogdill to get failed payment data through grep to Create tools to retrieve relevant information from logs in a useful format.Sep 24 2015, 2:29 AM
Ejegg updated the task description. (Show Details)
CCogdill_WMF raised the priority of this task from Low to Medium.Jan 19 2016, 7:17 PM

Bumping up priority now that we're out of Q2. The early tests we did of this kind of email suggests we are leaving money on the table by not following up with donors. I would love to have access to failed payment data sooner than Q4 if at all possible.

SBurnett-WMF subscribed.

No activity in nearly 10 years. Advised no longer being pursued.