Bug 1106656

Summary:	Docker run --memory value doesn't match cgroup value
Product:	Red Hat Enterprise Linux 7	Reporter:	Chris Evich <cevich>
Component:	docker	Assignee:	Matthew Heon <mheon>
Status:	CLOSED ERRATA	QA Contact:	Virtualization Bugs <virt-bugs>
Severity:	low	Docs Contact:
Priority:	unspecified
Version:	7.0	CC:	bsarathy, dwalsh, gouyang, jrieden, lsu, sghosh
Target Milestone:	rc	Keywords:	Documentation
Target Release:	---
Hardware:	x86_64
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2014-09-18 20:45:44 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	1109938

Description Chris Evich 2014-06-09 17:00:46 UTC

Description of problem:
Some values to the --memory option result in an exact match to the cgroup value, others are slightly off.  If expected, then this could be fixed with a documentation update.

Version-Release number of selected component (if applicable):
docker-0.11.1-10.el7.x86_64

How reproducible:
very

Steps to Reproduce:

docker run -d -t -m 524889 some_image bash
...
cat /sys/fs/cgroup/memory/system.slice/.../memory.limit_in_bytes 
5246976

Same result for --memory 524889 value suffixes of:
b and B  => cgroup: 5246976
k and K  => cgroup: 5368721408

Suffixes for 'm' and 'g' seem to correctly represent specified value.

Actual results:
Values specified on command line sometimes don't match value in cgroup file

Expected results:
Values from command line should match value in cgroup file, a message should
be printed or logged if the value is altered, and/or value manipulation/conformity rules should be documented for the --memory option.

Comment 1 Guohua Ouyang 2014-06-10 01:56:53 UTC

I think I caught this, the memory size must be a multiper of 4096.

I tested below:
set memory limits    crgoup memory limits
=================    ========================
1024m                1073741824
1g                   1073741824
1048576k             1073741824
1073741824           1073741824
1048576              1048576
1024575              1048576
1048577              1052672 (1048576 + 4096)
1052673              1056768 (1052672 + 4096)
=================    ========================

so this should be not a bug.

Comment 2 Guohua Ouyang 2014-06-10 02:13:41 UTC

for the example in the bug, 524889 is between 524288 ~ 528384, so the value in cgroups should be the later.

#docker run -d -t -m 524889 some_image bash

# cat docker-dd95e808b318c29f3f79266118768d4cd196f3a9f2eea88bc8d473a71b58b27d.scope/memory.limit_in_bytes 
528384

>>>528384 - 524288 (1024*4*(524889 / 4096)
4096

Comment 3 Chris Evich 2014-06-10 14:59:48 UTC

Great, thanks for confirming.  Moving this to 'low' severity then since it'll just be a short documentation/man page update then.  I'd suggest something like:

The value may be rounded up to the next multiple of the host's memory page size(?) (typically 4k) unless an exact multiple is given.

Comment 4 Matthew Heon 2014-06-25 16:47:59 UTC

Patch sent to upstream for this (https://github.com/dotcloud/docker/pull/6680). Should appear in our builds of docker-1.0 for RHEL7 soon.

Comment 5 Daniel Walsh 2014-06-27 16:26:44 UTC

Fixed in docker-1.0

Comment 10 errata-xmlrpc 2014-09-18 20:45:44 UTC

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHBA-2014-1266.html