Re: [PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-22 Thread Jan Kara
On Mon 21-10-19 23:49:04, Roman Gushchin wrote: > On Wed, Oct 16, 2019 at 11:18:40AM +0200, Jan Kara wrote: > > On Tue 15-10-19 21:40:45, Roman Gushchin wrote: > > > On Tue, Oct 15, 2019 at 11:09:33AM +0200, Jan Kara wrote: > > > > On Thu 10-10-19 16:40:36, Roman Gushchin wrote: > > > > > > > > >

Re: [PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-21 Thread Roman Gushchin
On Wed, Oct 16, 2019 at 11:18:40AM +0200, Jan Kara wrote: > On Tue 15-10-19 21:40:45, Roman Gushchin wrote: > > On Tue, Oct 15, 2019 at 11:09:33AM +0200, Jan Kara wrote: > > > On Thu 10-10-19 16:40:36, Roman Gushchin wrote: > > > > > > > @@ -426,7 +431,7 @@ static void inode_switch_wbs_work_fn(str

Re: [PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-16 Thread Jan Kara
On Tue 15-10-19 21:40:45, Roman Gushchin wrote: > On Tue, Oct 15, 2019 at 11:09:33AM +0200, Jan Kara wrote: > > On Thu 10-10-19 16:40:36, Roman Gushchin wrote: > > > > > @@ -426,7 +431,7 @@ static void inode_switch_wbs_work_fn(struct > > > work_struct *work) > > > if (!list_empty(&inode->i_io_l

Re: [PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-15 Thread Roman Gushchin
On Tue, Oct 15, 2019 at 11:09:33AM +0200, Jan Kara wrote: > On Thu 10-10-19 16:40:36, Roman Gushchin wrote: > > We've noticed that the number of dying cgroups on our production hosts > > tends to grow with the uptime. This time it's caused by the writeback > > code. > > > > An inode which is getti

Re: [PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-15 Thread Jan Kara
On Thu 10-10-19 16:40:36, Roman Gushchin wrote: > We've noticed that the number of dying cgroups on our production hosts > tends to grow with the uptime. This time it's caused by the writeback > code. > > An inode which is getting dirty for the first time is associated > with the wb structure (loo

[PATCH v2] cgroup, blkcg: prevent dirty inodes to pin dying memory cgroups

2019-10-10 Thread Roman Gushchin
We've noticed that the number of dying cgroups on our production hosts tends to grow with the uptime. This time it's caused by the writeback code. An inode which is getting dirty for the first time is associated with the wb structure (look at __inode_attach_wb()). It can later be switched to anoth