Bug 1794168 - net-snmp memory leak
Summary: net-snmp memory leak
Alias: None
Product: Red Hat Enterprise Linux 7
Classification: Red Hat
Component: net-snmp
Version: 7.7
Hardware: Unspecified
OS: Unspecified
Target Milestone: rc
: ---
Assignee: Josef Ridky
QA Contact: BaseOS QE - Apps
Depends On:
TreeView+ depends on / blocked
Reported: 2020-01-22 19:58 UTC by bofo13
Modified: 2020-01-28 11:12 UTC (History)
1 user (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Last Closed: 2020-01-28 09:26:48 UTC
Target Upstream Version:

Attachments (Terms of Use)

Description bofo13 2020-01-22 19:58:32 UTC
snmpd process from net-snmp gets bigger and bigger while it is requested.
Especially when using SNMPv3, SHA and AES.
On some servers process was using more than 500MB of memory, had to restart it to come back to the usual 10MB.
But process then grows again.
Could you look at this please ?
Thank you very much !

# uname -r
# cat /etc/centos-release
CentOS Linux release 7.7.1908 (Core)
# yum list installed | grep snmp
net-snmp.x86_64 1:5.7.2-43.el7 @base
net-snmp-agent-libs.x86_64 1:5.7.2-43.el7 @base
net-snmp-libs.x86_64 1:5.7.2-43.el7 @base
net-snmp-perl.x86_64 1:5.7.2-43.el7 @base
net-snmp-utils.x86_64 1:5.7.2-43.el7 @base

Initial report : https://bugs.centos.org/view.php?id=16927

Comment 2 bofo13 2020-01-22 20:00:34 UTC
Some kind related to bug #1779609.

Comment 3 Josef Ridky 2020-01-23 07:48:16 UTC
Bug #1779609 is about sth else, but I'll try to take a look at it.

Do you have any reproducer (e.g. content of /etc/snmp/snmpd.conf, used commands for running snmp queries and so on)?

Comment 4 bofo13 2020-01-23 10:59:54 UTC
Thank you Josef for your answer.

Here's the daemon config :

# grep -vE '^#.*|^$' /etc/snmp/snmpd.conf 
rouser r
syslocation Unknown (edit /etc/snmp/snmpd.conf)
syscontact Root <root@localhost> (configure /etc/snmp/snmp.local.conf)
dontLogTCPWrappersConnects yes
agentAddress udp:161
includeAllDisks 0
extend-sh testcmd echo OK

# grep -vE '^#.*|^$' /var/lib/net-snmp/snmpd.conf 
usmUser 1 3 0x80001f8880cf894745b6369a5c00000000 "r" "r" NULL . 0x4cd9dcb044a50668ebb54f73e81cf10ce41e77b1 . 0x4cd9dcb044a50668ebb54f73e81cf10c 0x
ifXTable .1 14:0 18:0x $
ifXTable .2 14:0 18:0x $
setserialno 1840552802
engineBoots 47
oldEngineID 0x80001f8880cf894745b6369a5c00000000

Command used to create the above user :
createUser r SHA "MyNicePassword" AES "MyNicePassword"

I request the following OID every minute :

And the following ones on a 3 minutes basis :
. / .

I don't know if it's related to a particular OID, or if it's a SNMP v3 / SHA / AES issue.
I would go for a SNMP v3 / SHA / AES issue, because using the exact same configuration with SNMP v2 does not trigger the issue.

May perhaps help, note that using the exact same configuration (OID, SNMP v3 / SHA / AES) on Debian 9/10 does not trigger the issue.

Feel free if needed, thank you !

Comment 5 Ties Voskamp 2020-01-25 12:59:30 UTC

I can confirm this issue. I see the memory usage (memory_rss) grow with every request that is handled bij net-snmpd.
It doesn't matter if I use the standard SNMP-OID's or use "extend" or "extend-sh".

When downgrading to net-snmpd 5.7.2-37 this behaviour stops.

If you need additional info to debug just let me know!

Kind regards,
Ties Voskamp

Comment 6 Josef Ridky 2020-01-28 09:26:48 UTC
After some investigation, I have good news.

This issue has been fixed in version net-snmp-5.7.2-45, which will be available in RHEL-7.8 release, that should be released in Q1 2020.

Only obstacle for me is, I don't know exactly, how long does it take to propagate RHEL-7.8 repositories to CentOS, but it's not aim of this bug report.

Due of fix availablility, I am closing this bug report as NEXTRELEASE.

Comment 7 bofo13 2020-01-28 11:12:01 UTC
Really good news, thank you very much Josef.

Note You need to log in before you can comment on or make changes to this bug.