On Thu, Nov 28, 2024 at 10:22:07AM +0100, Martin Pieuchot wrote: Hello Martin,
>> It's hard to tell. It looks like your machine somehow ended up without >> free DMA pages a situation from which the pagedaemon hardly recovers. >> >> If this happens again, wait a couple of minutes because the pagedaemon >> might take some time to recover. If the kernel doesn't recover, could >> you try to enter ddb(4) and show the content of "ps /o", "show uvm" and >> "show bcstats". > During h2k24 I did a lot of work to prevent the situation you ran into > and some more to recover from it. > > Please let us know if you still encounter something fishy. Thanks -- I've been following the changes and updating my kernel/snapshot regularly. So far I've not seen a recurrence of the problem since I saw your changes go in, which is very promising. Admittedly, I'm not quite sure what the trigger was, but I have been hammering that box in recent days, so far without consequence! Thanks for all your hard work -- it is much appreciated! Laurie
