https://crash-stats.mozilla.com/search/?product=Firefox&uptime=%3C8&safe_mode=__true__&process_type=%21plugin&process_type=%21content&signature=%3DOOM%20%7C%20small&_facets=signature&_facets=moz_crash_reason&_facets=abort_message&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform&_columns=last_crash&_columns=available_virtual_memory&_columns=available_physical_memory&_columns=available_page_file#crash-reports This is a list of crashes in OOM | small with very low uptime, and running in safe mode. I cannot think of a reason in normal operation why we'd run out of memory in this configuration. I think we need to investigate this further and break this down: * how many of these are actually running out of: ** physical memory ** actually running out of VM ** fragmenting VM to death ** not actually running out of memory - if so, why are we reporting this as OOM | small?
Possibly related (from an email thread about a surge in startup crash reports)... "Yes, it seems socorro is mis-classifying these. Are we using the uptime at the time of submission to classify start up crashes? If so, that's pretty broken, we should be using the uptime at the time of crashing."