It could be related. If a device cannot flush buffers to disk it may wind up just holding them in memory. That would be defective behavior though.
What is your storage subsystem made up of? Sent from my iPhone > On Oct 20, 2015, at 8:36 AM, Richard PALO <rich...@netbsd.org> wrote: > > Le 20/10/15 06:23, Robert Mustacchi a écrit : >> On 10/19/15 11:27 , Yuri Pankov wrote: >>> Yes, and it's actually a feature, not a bug - >>> https://www.illumos.org/issues/3246 >> >> To phrase this more practically, this means that some I/O that was >> issued was not returned. This could be because of a hardware or software >> bug. Figuring out what issued I/Os it was and to what device would be a >> useful next step to make forward progress here. I would start looking at >> the outstanding zio's in mdb with, IIRC something like ::walk zio | >> ::zio -r. >> >> Robert > > Okay, thanks.. I'm trying to make room to expand my dump correctly. > > In the meanwhile, I've noticed something interesting... > When I boot up, swap -hs tells me I have about 42G available > but after a period of time: >> richard@omnis:/home/richard$ swap -lh >> swapfile dev swaplo blocks free >> /dev/zvol/dsk/rpool/swap 90,2 4K 2,0G 2,0G >> /dev/zvol/dsk/dpool/swap 90,3 4K 16G 16G >> richard@omnis:/home/richard$ swap -hs >> total: 4,0G allocated + 290M reserved = 4,3G used, 16G available > > nothing from my 32G of RAM are available any longer, the kernel taking over > 50%. > attached is gzip'd screenlog with a ::memstat and a ::kmastat from 'mdb -k' > > perhaps related to the issue... > > -- > Richard PALO > > <screenlog.0.gz> ------------------------------------------- illumos-discuss Archives: https://www.listbox.com/member/archive/182180/=now RSS Feed: https://www.listbox.com/member/archive/rss/182180/21175430-2e6923be Modify Your Subscription: https://www.listbox.com/member/?member_id=21175430&id_secret=21175430-6a77cda4 Powered by Listbox: http://www.listbox.com
