Record histograms when an urgent discard request is received.
This CL removes the following histograms which were recorded before each individual tab discard: - Tabs.Discard.TabCount - Tabs.Discard.InitialTime2 - Tabs.Discard.IntervalTime2 And replaces them with histograms recorded when an urgent discard request is received, before any tab is discarded: - Discarding.Urgent.NumAliveTabs - Discarding.Urgent.TimeSinceStartup - Discarding.Urgent.TimeSinceLastUrgent Also, metrics::RecordMemoryStats() is called when an urgent discard request is received rather than before each individual tab discard. The reason for this change is that we are interested in knowing how often urgent discard requests occur and what leads to them. We don't care how often we proactively discard tabs and we don't need multiple histogram samples when we discard multiple tabs in response to a single discard request. Bug: 775644 Change-Id: Ic603a904dc9635b12b3204c1e12bc293c4dcbaf7 Reviewed-on: https://chromium-review.googlesource.com/786295Reviewed-by:Gayane Petrosyan <gayane@chromium.org> Reviewed-by:
Chris Hamilton <chrisha@chromium.org> Commit-Queue: François Doray <fdoray@chromium.org> Cr-Commit-Position: refs/heads/master@{#521389}
Showing
Please register or sign in to comment