nlplug-findfs timeout too small for some systems
I have an ASUS X99 extreme11, which has something like 18 SATA ports and a whole lot of other interfaces + PCIe cards.
My disk setup is UEFI, luks2-encrypted LVM, and then in the lvm boot and root partitions. This setup works fine on other systems like my laptop and another desktop, both amd64. This setup does not work with a system with a ton of i/o.
The system that has the X99 extreme gives me 1652 directories in /sys/devices/LNXSYSTM:00
alone, always leading nlplugfs to timeout and thus lead to a failure to boot:
nlplug-findfs: exit due to timeout (6000)
nlplug-findfs: modaliases: 272, forks: 0, events: 1044, total bufsize: 214523
The --timeout
variable to nlplugfs only affects the uevent timeout, and not the MAX_EVENT_TIMEOUT, so nlplug times out at 6000ms before there's a chance to process all uevents.
Would there be any objection to either increasing the MAX_EVENT_TIMEOUT?
If so, how about allowing MAX_EVENT_TIMEOUT in nlplug-findfs to be set from an argument (eg. -x, --event-timeout TIMEOUT timeout after TIMEOUT milliseconds
), and then this argument can be set via mkinitfs (eg. EVENT_TIMEOUT=TIMEOUT
)?