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
Debugging with @matbryan52: if the Merlin dataset has many files (thousands), it will exhaust the number of file descriptors available by default. Increasing the limit can then cause another issue, as we are using select to check for connection abortions, and select only supports file descriptors in [0, 1024).
Why we are using select: Windows. See commit ec9953c
Possible fix is to keep using select on Windows, and switch to the previous implementation using poll or epoll for Linux, which works properly for "high" file descriptor numbers.
The text was updated successfully, but these errors were encountered:
Debugging with @matbryan52: if the Merlin dataset has many files (thousands), it will exhaust the number of file descriptors available by default. Increasing the limit can then cause another issue, as we are using
select
to check for connection abortions, andselect
only supports file descriptors in[0, 1024)
.Why we are using
select
: Windows. See commit ec9953cPossible fix is to keep using
select
on Windows, and switch to the previous implementation usingpoll
orepoll
for Linux, which works properly for "high" file descriptor numbers.The text was updated successfully, but these errors were encountered: