You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Observed: When copy_proc_mem fails (i.e. both rv != 0 and the context's depth < 0 ) for the first function on the stack trace, it sleeps for --sleep-ms seconds
Expected: There should be an option to aggressively retry printing the stack from the first step immediately without sleeping, on failure, so that the phpspy output reflects a running application's stack trace accurately, and has traces from as many milliseconds as possible. (Optionally suppress the repeated warnings until the time elapses or the first success)
Alternative: Print a suggestion to use --pause-process when copy_proc_mem fails if that option is not set, mentioning that it is not safe in production
How about something like --immediate-retry=<n> where n is the number of times to retry or -1 to retry indefinitely? -1 could be unsafe as it could peg CPU if memory cannot be read for some reason. Of course we should never retry if the pid is detected to be dead.
See #111 (comment) for context
Observed: When
copy_proc_mem
fails (i.e. both rv != 0 and the context's depth < 0 ) for the first function on the stack trace, it sleeps for--sleep-ms
secondsExpected: There should be an option to aggressively retry printing the stack from the first step immediately without sleeping, on failure, so that the phpspy output reflects a running application's stack trace accurately, and has traces from as many milliseconds as possible. (Optionally suppress the repeated warnings until the time elapses or the first success)
Alternative: Print a suggestion to use
--pause-process
when copy_proc_mem fails if that option is not set, mentioning that it is not safe in productionWorkarounds
https://github.com/adsr/phpspy#demos doesn't mention
--pause-process
in the instructions for profiling, but the help text in the README does.The text was updated successfully, but these errors were encountered: