Hi,

Thanks for your response. I've answered your questions as well as I could
and included a "watch" showing the memory leak.


> Which version of the agent are you using?
>

I am compiling from the package 5.6.2 pre 1

> Is this an agent that you compiled yourself, or a vendor-supplied version?
>
>

I am compiling it myself.  Cross compiled for ARM on an i686 linux VM. I
have not yet added my own modules. Here is the config.

./configure  --enable-mini-agent \
--with-mib-modules="disman/event ucd_snmp ucd-snmp/proc ucd-snmp/loadave
ucd-snmp/memory mibII mibII/sysORTable mibII/system_mib if-mib ip-mib
tcp-mib udp-mib" \
--enable-static --disable-shared \
--prefix=`pwd`/install --exec_prefix=`pwd`/install-arm \
--host=arm-none-linux-gnueabi --target=arm-none-linux-gnueabi
--build=i686-linux \
--with-cc=arm-none-linux-gnueabi-gcc \
--with-linkcc=arm-none-linux-gnueabi-gcc \
--with-ld=arm-none-linux-gnueabi-ld \
--with-ar=arm-none-linux-gnueabi-ar \
--with-sys-contact="[email protected]" \
--with-sys-location="ip-pqube" --with-default-snmp-version="2" \
--with-logfile="/var/log/snmpd.log" \
--with-persistent-directory="/var/net-snmp" \
--without-rpm --disable-embedded-perl --disable-deprecated
--disable-manuals --disable-ipv6

Here is the config summary.  The only message I see otherwise is the one
about enabling sysctl.

---------------------------------------------------------
            Net-SNMP configuration summary:
---------------------------------------------------------

  SNMP Versions Supported:    1 2c 3
  Building for:               linux
  Net-SNMP Version:           5.6.2.pre1
  Network transport support:  Callback Unix Alias TCP UDP IPv4Base
SocketBase TCPBase UDPIPv4Base UDPBase
  SNMPv3 Security Modules:     usm
  Agent MIB code:             disman/event ucd_snmp ucd-snmp/proc
ucd-snmp/loadave ucd-snmp/memory mibII mibII/sysORTable mibII/system_mib
if-mib ip-mib tcp-mib udp-mib default_modules => 
disman/event/mteScalars disman/event/mteTrigger
disman/event/mteTriggerTable disman/event/mteTriggerDeltaTable
disman/event/mteTriggerExistenceTable
disman/event/mteTriggerBooleanTable
disman/event/mteTriggerThresholdTable disman/event/mteTriggerConf
disman/event/mteEvent disman/event/mteEventTable
disman/event/mteEventSetTable disman/event/mteEventNotificationTable
disman/event/mteEventConf disman/event/mteObjects
disman/event/mteObjectsTable disman/event/mteObjectsConf ucd-snmp/disk
ucd-snmp/proc ucd-snmp/versioninfo ucd-snmp/pass ucd-snmp/pass_persist
ucd-snmp/loadave agent/extend ucd-snmp/errormib ucd-snmp/file
ucd-snmp/dlmod ucd-snmp/proxy ucd-snmp/logmatch ucd-snmp/memory
ucd-snmp/vmstat util_funcs util_funcs/header_simple_table
hardware/memory mibII/snmp_mib_5_5 mibII/system_mib mibII/sysORTable
mibII/at mibII/ifTable mibII/ip mibII/tcp mibII/icmp mibII/udp
mibII/vacm_vars mibII/setSerialNo if-mib ip-mib ip-forward-mib tcp-mib
udp-mib util_funcs mibII/updates if-mib/ifTable if-mib/ifXTable
ip-mib/ipAddressTable ip-mib/ipAddressPrefixTable
ip-mib/ipDefaultRouterTable ip-mib/inetNetToMediaTable
ip-mib/ipSystemStatsTable ip-mib/ip_scalars
ip-mib/ipv6ScopeZoneIndexTable ip-mib/ipIfStatsTable
tcp-mib/tcpConnectionTable tcp-mib/tcpListenerTable
udp-mib/udpEndpointTable snmpv3mibs mibII/snmp_mib mibII/system_mib
mibII/sysORTable mibII/vacm_vars mibII/vacm_conf util_funcs/Exit
util_funcs/restart
  MYSQL Trap Logging:         unavailable
  Embedded Perl support:      disabled
  SNMP Perl modules:          disabled
  SNMP Python modules:        disabled
  Crypto support from:        internal
  Authentication support:     MD5 SHA1
  Encryption support:         DES AES

---------------------------------------------------------


The ensuing make is squeaky clean.


>> Every few seconds, and monitoring the process I see the size of the
>> daemon
>> go up 600-1000k.
>
> Hmmm....  that's an incredibly large jump.

Yes but it is the case.  The RAM disappears and the system crashes after a
minute or so.  If I see the vacm message--which I see when I uncomment
that line--it happens.  If I don't, it does not. Here is an example:

Commented out monitor line:

# watch -t -n 1 ps -o comm,vsz,rss | grep snmpd
snmpd            6120 3380
snmpd            6120 3380
snmpd            6120 3380
snmpd            6120 3380
snmpd            6120 3380
snmpd            6120 3380
snmpd            6120 3380
...

# kill -HUP 640 (hup edited config to uncomment the offending line)
# watch -t -n 1 ps -o comm,vsz,rss | grep snmpd
snmpd            6368 3656
snmpd            6428 3780
snmpd            6476 3808
snmpd            6560 3920
snmpd            6680 4084
snmpd            6800 4208
snmpd            6920 4336
snmpd            7040 4460
snmpd            7220 4584
snmpd            7340 4708
snmpd            7460 4828
snmpd            7520 4932
snmpd            7640 5060
snmpd            7820 5184
snmpd            7940 5308
snmpd            8000 5420
snmpd            8180 5548
snmpd            8300 5672
snmpd            8420 5800
snmpd            8540 5924
snmpd            8660 6048
snmpd            8780 6172
snmpd            8900 6296
snmpd            9020 6420
snmpd            9140 6548
snmpd            9260 6648

# kill -9 640


> I know there were minor memory leaks in some versions of the software,
> but nothing of that scale.
>

>> Here is my access control:
>>
>> rocommunity public
>> rwcommunity private
>
> You may want to set a different "rwcommunity".
> Letting the whole world know how to manipulate your systems is,
> how can I put this...,   unwise in the extreme!
>
>

Ha! Yes, as I wrote above, I suspected that the acm config was busted, and
a lot of what is in it now are "throw it at the machine" attempts to see
if I could fix the leak.  The "rouser" was the last thing I tried, and
wasn't there for the previous attempts.  Adding it (or not) makes no
difference.

> So this configuration is inconsistent.
>
>

Yupp. I will try making it consistent, but as it is it crashes the device.

Thanks! (Again.)

D.



------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Net-snmp-users mailing list
[email protected]
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users

Reply via email to