- 1 Meeting Info
- 2 Operations Updates
- 3 Project Updates
- 4 Major Projects
- 4.1 Migrating off of persona
- 4.2 Sending public data to parquet for reading from spark/re:dash
- 4.3 Symbols service refactoring (snappy, somewhat tangental to us)
- 4.4 Signature generation across crash reporters
- 4.5 Splitting out collector
- 4.7 Handling more PII data in crashes
- 4.8 Sending stacks for all crashes from the client
- 4.9 Replacing FTPscraper
- 5 other business
- 6 Travel, etc
- 7 Links
Breakpad status meetings occur on Wed at 10:00am Pacific Time.
Vidyo: Stability 650-903-0800 x92 conf 98200# 800-707-2533 (pin 369) conf 98200#
IRC backchannel: #breakpad
Mountain View: Dancing Baby (3rd floor)
- Stage is down :(
- jp will fix today.
- Thanks willkg! https://bugzilla.mozilla.org/buglist.cgi?product=Socorro&component=Infra&resolution=---
- Stage submitter
- needs more love for the future https://bugzilla.mozilla.org/show_bug.cgi?id=1289466
- Pingdom account
- fixed for now on existing account
- jp will set up new user accounts under Travis's payment plan
- ElasticSearch monitoring?
- Not going to use DataDog
- jp will investigate some new possible tools
- Jason on service-ops team is going to help.
- What's up Ruby 2.0 on Travis?
- Travis vs Leeroy?
- Can we move or resolve ALL bugs in Socorro::Middleware? (**Yes!**)
- Perhaps do the same with Socorro::Database
- peterbe resolved the last bug in Socorro::Bixie and asked bmo if they can shut down that component
Migrating off of persona
- adrian has r+'d the pull request
- user experience regresses because it forces sign in everyday
- will land on stage shortly
- after validation, we will start communicating with the staff and active non-staff
Sending public data to parquet for reading from spark/re:dash
- peterbe met with Mark Reid,
- Plan is to add another thing to our PolyCrashStorage
- Upload a cut of the processed crash with ONLY the fields in the processed crash JSON Schema
- Write some Scala code that iterates all the keys that we have in our JSON Schema
- Still need to think about reprocessing
- We will not be sending directly to their ingestion servers (nginx/heka)
Symbols service refactoring (snappy, somewhat tangental to us)
- Gabriele is back from PTO and mentioned redoing it in Rust.
- It's not clear what the requirements are, so Will is passively monitoring this.
- Talked to rweiss about doing some classification or other statistical correlation and the modelling is tricky and probably needs a specialist. We have a whole team of applied mathematicians working in 'metrics' that could help, but this is not something we will get to before Hawaii. [lonnen]
Signature generation across crash reporters
- don't know what the next step should be, on hold [Adrian]
- current prototype: https://github.com/adngdb/crash-signature-service - https://crash-signature-service.herokuapp.com/sign (see README in github)
- waiting for a non-socorro project to need it
- the fuzzmanager project/needs are very different from what we do/have
Splitting out collector
- Will talked with Lonnen about life and ops and infrastructure and is in the process of moving the MVP line and rethinking parts of the project. Taking stock of things. More on this next week.
- Antenna is back!
- JP could move it today, but Travis's deployment patterns are very different from ours
- JP has set up a meeting with Lonnen with Will to sort out the process and features of moving.
- slipping out of this Q
- No updates
- Lonnen to chase brad lassey
Handling more PII data in crashes
- deferring work for now
- needs coordination with the data platform team
Sending stacks for all crashes from the client
- no new updates since last week
- [lonnen] to talk with ddurst about this before next week
- peterbe met with Michael Shal
- Notes: https://public.etherpad-mozilla.org/p/socorro-releng-index-201607
- Reduced down to a couple of important questions about whether the TaskCluster Index has all the right stuff
- and if it does, which exact locations and .json files should we pull down.