For ANRs, reduce EXPIRED_REPORT_TIME_MS to 10s
After 10s, the state of other processes is likely to not be useful anyway to debug the ANR. If there is already a delay on the queue, not dump all the proceses will also help to recover more quickly so that future ANR snapshots are more accurate (smaller latency) Change-Id: Ibc4c85d90389724a0bc214f789f64edebb59da78 Bug: 254634348 Test: unit tests
This commit is contained in:
@@ -49,7 +49,7 @@ class AnrHelper {
|
||||
* this time, the information might be outdated. So we only the dump the unresponsive process
|
||||
* instead of including other processes to avoid making the system more busy.
|
||||
*/
|
||||
private static final long EXPIRED_REPORT_TIME_MS = TimeUnit.MINUTES.toMillis(1);
|
||||
private static final long EXPIRED_REPORT_TIME_MS = TimeUnit.SECONDS.toMillis(10);
|
||||
|
||||
/**
|
||||
* If the last ANR occurred within this given time, consider it's anomaly.
|
||||
|
||||
Reference in New Issue
Block a user