Re: [Devel] [PATCH rh7 04/11] ub: do not apply memcg limits on attaching via sys_setluid

Kirill, please review. -- Best regards, Konstantin Khorenko, Virtuozzo Linux Kernel Team On 05/20/2015 06:49 PM, Vladimir Davydov wrote: I recalled why I had added this patch - we have to activate kmem in the underlying memcg before attaching a task to a beancounter. However, it is

Re: [Devel] [PATCH rh7 0/4] ub: fix memcg resource accounting

В Вт, 19/05/2015 в 17:31 +0300, Vladimir Davydov пишет: UB_PHYSPAGES, UB_SWAPPAGES, UB_KMEMSIZE are now accounted on the memcg size. However, some stats differ between beancounters and memcg (e.g. failcnt). This patch set fixes them. For more details, see individual patches. Related to

[Devel] [PATCH RHEL7 COMMIT] bc/memcg: show correct swap max for beancounters

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit a74376e2dde136629060dcb957d84c85545e3961 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 13:18:53 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] bc/memcg: show correct mem/swap failcnt for beancounters

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 3fe10e41f46d0078bf6fb50a98d5c1becd5c2b82 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 13:19:04 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] bc: deprecate kmemsize limit

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 89790f478519e4b319446fb9ebc2d50a5141034e Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 13:19:16 2015 +0400

[Devel] [PATCH rh7 0/2] fairsched stats

Currently, cpu stats are broken for UUID-named CTs. First, reading /proc/stat from inside a container or /proc/vz/vestat on the host fails with ENOENT if there is a UUID-named CT. Patch 1 fixes that. Second, there is no new interface for getting loadavg for UUID-named CTs, which would be a

[Devel] [PATCH RHEL7 COMMIT] bc/memcg: assure swap-held/maxheld swap-limit for beancounters

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 6a541e8212b0bffc602a3fb998a04a5484b21c53 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 13:19:10 2015 +0400

[Devel] [PATCH rh7 1/2] farisched: switch methods showing stats from numeric id to name

farisched_get_cpu_stat, fairsched_get_cpu_avenrun, and farisched_show_stat can be called on UUID-named containers (e.g. reading /proc/stat inside a CT or reading /proc/vz/vestat on the host). Currently, they return ENOENT for such containers, because they try to open cgroup with name equal to the

[Devel] Running Debian 8 Jessie as OpenVZ Hostnode?

Hi, I've not been successful trying to install OpenVZ on the current stable version of Debian 8 Jessie. I was able to boot the OpenVZ stab108 kernel on Debian 8 Jessie with SysVinit. However I could not start a container. First it gave an error about not finding a kernel module. So I

Re: [Devel] [RFC rh7] net: venet -- Cleanup ip address on ve exit

On Wed, May 20, 2015 at 08:21:59PM +0300, Cyrill Gorcunov wrote: While been playing with c/r of a container with IP assigned I found that VE exit (ve_drop_context) is happening earlier than venet::exit routine which means the ve::ve_netns = nil when we enter into ip releasing procedure. |

[Devel] [PATCH RH7] idr: print a stack dump after ida_remove warning

backport of mainline commit dd04b452f532ca100f7c557295ffcbc049c77171 Author: Jean Delvare jdelv...@suse.de Date: Wed Jul 3 15:08:47 2013 -0700 idr: print a stack dump after ida_remove warning We print a dump stack after idr_remove warning. This is useful to find the faulty piece

[Devel] [PATCH RHEL7 COMMIT] ub: do not apply memcg limits on attaching via sys_setluid

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 21981083cfc6eefc4f110dcd556fdf235800d711 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:27:45 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: sync memcg before setting limit via sys_setublimit

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit c613b7dff258b96fbc014613dc4a0a95eae18a66 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:27:58 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: get rid of pincount

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 7bf553f754a24a2ce21971bc5b94daa10597e538 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:28:44 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: zap /proc/bc/ID/slabinfo

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 07a8568b4fcfad418f7b8487fce524088bccc5b0 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:28:58 2015 +0400

[Devel] [PATCH rh7 2/2] fairsched: add cpu.proc.loadavg to the cpu cgroup

This file mimics behavior of the system wide /proc/loadavg. It is going to be a successor of the legacy VZCTL_GET_CPU_STAT ioctl. Also, use it for showing /proc/loadavg inside a container, just like we do in case of /proc/stat. https://jira.sw.ru/browse/PSBM-32284 Signed-off-by: Vladimir

Re: [Devel] [RFC rh7] net: venet -- Cleanup ip address on ve exit

On Wed, May 20, 2015 at 08:21:59PM +0300, Cyrill Gorcunov wrote: While been playing with c/r of a container with IP assigned I found that VE exit (ve_drop_context) is happening earlier than venet::exit routine which means the ve::ve_netns = nil when we enter into ip releasing procedure. |

[Devel] [PATCH RHEL7 COMMIT] ub: remove some dead code

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 2cd3d6168e3e3e76d0c3c8d6f501c73b56332154 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:26:40 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: simplify memcg synchronization

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit ae16e7f436ae461a50a65f05576109fdad70dbb4 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:27:27 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: kill ub_dcache_threshold sysctl knob

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 235e9340f66d8b55fec890148f579f2f18105d33 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:28:24 2015 +0400

Re: [Devel] [PATCH rh7 02/11] ub: zap show_ub_mem

On 22.05.2015 22:07, Konstantin Khorenko wrote: On 05/20/2015 12:00 PM, Vladimir Davydov wrote: It is called by the showmem sysctl. Hell knows why it was added and who would ever used it. It's not about sysctl, but about sysrq and we used it heavily until we start getting crash dumps from

[Devel] Any plans about overlayfs support for Docker?

Hello, folks! I inspired with Docker support inside containers so much! It works perfectly. But I can't find any articles regarding to OverlayFS support there. Do you have any plans about OverlayFS http://blog.cloud66.com/docker-with-overlayfs-first-impression/? It's must feature for Docker

[Devel] [PATCH RHEL7 COMMIT] ub: zap show_ub_mem in Alt+SysRQ+m

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 7af2449e25e83f47d87ccdfa516fefc96962d38c Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:27:08 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: drop ubc_ioprio knob

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 5bdcc62ebadc6934a6a360eaa57588baa0aff282 Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:28:12 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: do not show dead beancounters under /proc/bc

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit 0edaafd2a02f07bad38cc6faffa55cdd5d95cebc Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:28:36 2015 +0400

[Devel] [PATCH RHEL7 COMMIT] ub: zap /proc/bc/ID/dcacheinfo

The commit is pushed to branch-rh7-3.10.0-123.1.2-ovz and will appear at https://src.openvz.org/scm/ovz/vzkernel.git after rh7-3.10.0-123.1.2.vz7.5.5 -- commit e2cc609ce6c3881c0f3474526b26637eba4e98ef Author: Vladimir Davydov vdavy...@parallels.com Date: Mon May 25 14:29:11 2015 +0400

[Devel] [PATCH rh7 1/3] memcg: rename memory.low_limit_in_bytes to memory.low

This is how it called upstream (in the unified hierarchy). Another reason for rename is that memory.low is not actually a limit - it is a low wss estimate. Since memory.low was added not long before and it is only used by vcmmd and we are haven't released Beta yet, such a rename seems to be

[Devel] [PATCH rh7 2/3] memcg: use CFTYPE_NOT_ON_ROOT for memory.low and memory.oom_guarantee

This is neater than checking if the root is passed to the write method and this is how it works upstream (for memory.low). Signed-off-by: Vladimir Davydov vdavy...@parallels.com --- mm/memcontrol.c | 8 ++-- 1 file changed, 2 insertions(+), 6 deletions(-) diff --git a/mm/memcontrol.c

[Devel] [PATCH rh7 3/3] memcg: port memory.high

This patch backports memory.high knob, which was introduced upstream by commit 241994ed8649 (mm: memcontrol: default hierarchy interface for memory): - memory.high configures the upper end of the cgroup's expected memory consumption range. A cgroup whose consumption grows beyond this

Re: [Devel] [PATCH rh7 1/3] memcg: rename memory.low_limit_in_bytes to memory.low

Kirill, please review this set. On Mon, May 25, 2015 at 05:05:46PM +0300, Vladimir Davydov wrote: This is how it called upstream (in the unified hierarchy). Another reason for rename is that memory.low is not actually a limit - it is a low wss estimate. Since memory.low was added not long