Breakpad/Status Meetings/2008-Dec-3

From MozillaWiki
Jump to: navigation, search

ideas

  • store user ids of reports selected manually for priority jobs, auto-process if seen again
  • link to crash report immediately in crash report dialog so people know how to see it right away

deferred job status

  • deferred jobs cleanup script wasn't working as expected
  • aravind wrote a secondary script to scan and delete deferred files older than a certain date
  • no rush to push updated cleanup script, have time to do it right
  • 1.5T available
    • 158G available by symbols
    • 375G left as of 1132am 12/3
  • at 75% of capacity, 375G left
  • before cleanup we were at 95% capacity
  • database requirements - 200G disk, 67%
    • can we archive partitions older than 120 days

deferred job space solutions

  • right way
    • client-side feasibility of gzipping in crash report client
    • can stackwalk be modified to accept/determine gzip vs. uncompressed
  • right but hacky way
    • gzipping in collector raises cpu load in the webheads
    • have stackwalk detect gzip and inflate as necessary
  • not so right and hacky way
    • gzipping in collector raises cpu load in the webheads
    • processor inflates before calling stackwalk
  • crappy way
    • have a secondary process gzipping deferred jobs, hoping there aren't collisions
    • processor tries/fails between compressed and uncompressed
  • get bigger disks
    • possible, but not really necessary
  • shorten deferred jobs lifetime
    • 12 days? 10 days?