CrashKill/2014-07-07

General

Previous Action Items

• [KaiRo] Ask relman if we can put a field into uplift requests to ask for prerequisites
• ashughes to update the group on GGC status when he's back from PTO

Socorro

• 16 minute outage to all PHX1 hosts Sat night, crash-stats missing data for that portion of the evening
• https://bugzilla.mozilla.org/buglist.cgi?bug_id=1018349%2C1030276%2C1030218%2C1021471%2C1030878%2C1017077%2C1031076%2C1019262%2C1014128%2C977634%2C1020348%2C979357%2C977636&list_id=10674597

Desktop

Beta

• bug 991845 (OOM in WorkerJSRuntime) is worsened by GGC, see comment #15. bug 1034611 is one part to help and has a patch, bug 1034621 is another very GGC-centric piece.
• bug 944873 (AsyncShutdown crashes) has multiple sources, but bug 1034726 should help us better determine those, and bug 1008944 and bug 1034975 fix fix some common cases.
  • That said, Yoric found out that experiments (bug 1012924) are the large issue on beta and the reason why this rose into topcrash ranges last week.
  • bug 1035144 is also filed to get AsyncShutdown into signatures as many of those crashes can be dataloss issues.

Aurora

• bug 1027103 (a Linux repro case of PushClipsToDT) has some analysis of what's going on and avenues to fix. bug 1034584 should fix the crash itself, but bug 1034593 goes more into the depth of the issue. This was reported against Aurora, but might affect beta as well.

Trunk

• New graphics crash: bug 1035168 - needs dev attention.

Android

• The NSS 3.16.0 to 3.16.2 upgrade or the PKIX changes in have removed the NSS crasher from the topcrash issues and may have removed the crash signature completely
• This has brought our crash rate from > 2.6 to near 2.0 on Beta
• aurora seems to be up from last week 5 crashes to 9
• nightly looks to be stable

Notes

Upcoming PTO

• Tracy - PTO through July 13