From: Andrey Ryabinin
Add new memcg file - memory.cache.limit_in_bytes.
Used to limit page cache usage in cgroup.
https://jira.sw.ru/browse/PSBM-77547
Signed-off-by: Andrey Ryabinin
khorenko@: usecase:
imagine a system service which anon memory you don't want to limit
(in our case it's a vSto
From: Andrey Ryabinin
Add new memcg file - memory.cache.limit_in_bytes.
Used to limit page cache usage in cgroup.
https://jira.sw.ru/browse/PSBM-77547
Signed-off-by: Andrey Ryabinin
khorenko@: usecase:
imagine a system service which anon memory you don't want to limit
(in our case it's a vSto
From: Valeriy Vdovin
Each VE should execute release agent notifications within it's own
workqueue. This way we achieve a more fine-grained control over
release_agent work flushing at VE destruction.
Short history of the patch:
https://jira.sw.ru/browse/PSBM-83887
Signed-off-by: Valeriy Vdovin
From: Valeriy Vdovin
Previous patches already switched release agent binary calling mechanism
to call the binary in ve. Now we need to switch the binary path to
per-ve path. So we store release agent path in ve_ra_data structures in
per-ve ve->ra_data_list.
Short history of the patch:
Signed-of
Patches are massively reworked, see inpatch comments.
https://jira.sw.ru/browse/PSBM-134002
Pavel Tikhomirov (1):
ve/cgroup: fix cgroup_mark_ve_roots naming
Valeriy Vdovin (5):
cgroup/cfs: added 'activate' option to cgroup_add_file
ve/cgroup: Implement per-ve workqueue
ve/cgroup: Move re
From: Valeriy Vdovin
In kernfs files get created in 'deactivated' state, which means they are
not visible. Add option to activate the file after creation immediately
making it visible in the parent directory. Will be used when adding
release agent files to ve cgroup roots.
Signed-off-by: Valeriy
Let's use proper English, plural nouns should have -s ending. We mark
roots for many container cgroups not for just one root. We already have
-s in "cgroup_unmark_ve_roots", let's be consistent.
https://jira.sw.ru/browse/PSBM-134002
Fixes: 907c1e6cd67f ("cgroup: Mark cgroup CGRP_VE_ROOT")
Signed-o
From: Valeriy Vdovin
To save release agent binary path per-ve we need to have some place to
store it, ve root cgroups are already linked to ve via cgroup->ve_owner
and ve->ve_ns->cgroup_ns->root_cset, so we can put release agent data to
ve. Also we need to have one path per cgroup_root as it prev
From: Valeriy Vdovin
Per-ve workqueue is started at ve cgroup start before ve->is_running is
set to true, it is stopped at ve cgroup stop after ve->is_running is set
to false and is guarded via synchronize_rcu, stop also implies waiting
for all queued and yet unfinished works to finish.
So the u
The assumption that bm_fill_super() is not called for the second time
for CT is wrong: umount operation clears sb->s_root, which causes
vfs_get_super() to call fill_super again on the next mount.
Make bm_fill_super() handle multiple-calls corrently:
- initialize bm_data and set ve->binfmt_misc onl
vz7/vz8/vz9?
--
Best regards,
Konstantin Khorenko,
Virtuozzo Linux Kernel Team
On 18.10.2021 19:51, Nikita Yushchenko wrote:
The assumption that bm_fill_super() is not called for the second time
for CT is wrong: umount operation clears sb->s_root, which causes
vfs_get_super() to call fill_supe
It shall be [PATCH vz8] - sorry for missing that.
vz7/vz8/vz9?
--
Best regards,
Konstantin Khorenko,
Virtuozzo Linux Kernel Team
On 18.10.2021 19:51, Nikita Yushchenko wrote:
The assumption that bm_fill_super() is not called for the second time
for CT is wrong: umount operation clears sb->s_
12 matches
Mail list logo
