Bug 1845418

Summary: The linux bridge `ageing_time` should not be mulitple by 100
Product: Red Hat Enterprise Linux 8 Reporter: Gris Ge <fge>
Component: NetworkManagerAssignee: sushil kulkarni <sukulkar>
Status: CLOSED NOTABUG QA Contact: Desktop QE <desktop-qa-list>
Severity: high Docs Contact:
Priority: unspecified    
Version: ---CC: acardace, atragler, bgalvani, lrintel, rkhan, sukulkar, thaller, till
Target Milestone: rc   
Target Release: 8.3   
Hardware: x86_64   
OS: All   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2020-06-09 12:10:48 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: 1738136    

Description Gris Ge 2020-06-09 07:54:19 UTC
Description of problem:


The sysfs already convert the jiffies into seconds before showing,
NM should not multiple it by 100.

Kernel code(br_sysfs_br.c) is:

static ssize_t ageing_time_show(struct device *d,
				struct device_attribute *attr, char *buf)
{
	struct net_bridge *br = to_bridge(d);
	return sprintf(buf, "%lu\n", jiffies_to_clock_t(br->ageing_time));
}


Version-Release number of selected component (if applicable):
NetworkManager-1.25.2-25985.d1e8eb791c.el8.x86_64

How reproducible:
100%

Steps to Reproduce:
1. sudo nmcli c add type bridge ifname br0
2. cat /sys/class/net/br0/bridge/ageing_time
3.

Actual results:

30000

Expected results:

300

Additional info:

Please also check with other bridge options on this `user_hz_compensate = TRUE`.
I believe all sysfs bridge option is converted to seconds.

Comment 1 Thomas Haller 2020-06-09 09:29:16 UTC
The bridge.ageing-time property in NetworkManager is in seconds (with a default of 300). It's also documented that way.

The sysctl value is in a different unit (1/100 of a second).

Of course, if NM configures a aging-time of 300 seconds, then the value on sysctl reads as "30000 1/100 seconds".

What is wrong with that?

Comment 2 Thomas Haller 2020-06-09 09:31:45 UTC
The problem here is that the setting in kernel has a wider range (supporting sub-seconds precision).
In NetworkManager API you currently cannot express such values.

The real problem is that this is public API, and it cannot be changed. It could only be deprecated and extended with a new value, which is a painful thing to do.

Comment 3 Gris Ge 2020-06-09 12:10:30 UTC
aha. My fault. It was multiple by the USER_HZ which could be retrieved via `os.sysconf("SC_CLK_TCK")` in python.