RE: snmpd crashed by snmpwalk

[Jim Su]

Title: RE: snmpd crashed by snmpwalk

Thanks John:

 

           I have few plan to avoid this problem.

           1. Comment out the code of "case IFPHYSADDRESS" as described in the bug report. But I don't know if there is any place "put 64 bit info into a 32 bit field".

        2. Take out the mibII table.

        3. Try net-snmp 5.1.2.

        4. Try ucd-snmp 4.2.5

 

        I'd appreciate if you share your experience if you have done that before or you have any good suggestion other than the above four.

       

 

Thanks,

Jim

-----Original Message-----
From: John Blake [mailto:[EMAIL PROTECTED]
Sent: Thursday, September 23, 2004 8:56 PM
To: 'Jim Su'; Net-Snmp-Users
Subject: RE: snmpd crashed by snmpwalk

 The actual issue is this;

 http://sourceforge.net/tracker/?group_id=12694&atid=112694&func=detail&aid=556830

 that's the bug Id for it.

 I'm not sure if it happens in Lunix or not.

 I am just a newbie at snmp and I don't know enough to mess with changing the code like in the link.

 It ahs to do with trying to put 64 bit info into a 32 bit field.

 

 John

 

-----Original Message-----
From: Jim Su [mailto:[EMAIL PROTECTED]
Sent: Thursday, September 23, 2004 12:23 AM
To: John Blake; Net-Snmp-Users
Subject: RE: snmpd crashed by snmpwalk

Hi John:

 

       Thanks for the input. I did some experiment on my target machine. I asked the owner of the target machine to check if there is any card has no IP address. He found one and added the IP address.

       The snmpwalk still crashed the system. The OS is Linux. I am going to build a net-snmp only snmpd to experiement that crashing is as you pointed. Do you know is there any patch available? Or if this is fixed in a latest version? Mine is 5.1.1, quite new. From your experience is 5.1.2 stable enough?

       I was told there is no such problem in ucd-4.2.5. I may back ward my application to ucd-4x. But there is some know issues in ucd-4x too. Any suggestion?

      Tx.

-Jim 

-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of John Blake
Sent: Wednesday, September 22, 2004 8:44 PM
To: 'Jim Su'; Net-Snmp-Users
Subject: RE: snmpd crashed by snmpwalk

 Jim,
 I had the same issue but it was because of a bug in SNMP. If there is no ip address on the interface, then it crashes. My OS is solaris.

 http://sourceforge.net/tracker/?group_id=12694&atid=112694&func=detail&aid=755477
 

 John
-----Original Message-----
From: Jim Su [mailto:[EMAIL PROTECTED]]
Sent: Tuesday, September 21, 2004 11:19 PM
To: Net-Snmp-Users
Subject: snmpd crashed by snmpwalk

Hi Greeting:

    I encountered one snmpd crashed due to the snmpwalk timeout at the "IF-MIB::ifNumber.0 ". All the values before IF-MIB were returned. But there is time out on IF-MIB.

    I ran the snmpwalk to another host. It went through okay. The two systems have different OS level of Linux. The snmpd was developed and build on "Linux  2.4.18" and then was

    installed on "Linux 2.4.21-mtd.1" All the snmpwalk on "Linux  2.4.18"  got returned values and didn't crash the snmpd.

    The system which was crashed by snmpwalk has OS  "Linux 2.4.21-mtd.1". 

    I have couple questions. If the OS level difference is critical then why the MIB values before IF-MIB are returned
    fine?  It should not return anything from the beginning, right? Could it be some specific device interface hard to be recognized on the target machine which causing the crash? 

   However the man page of "snmpwalk" said -

       "If  no  OID argument is present, snmpwalk will search MIB-2.  If the network entity has an error processing the
       request packet, an error packet will be returned and a message will be  shown,  helping  to  pinpoint  why  the
       request was malformed.

       If  the  tree search causes attempts to search beyond the end of the MIB, the message "End of MIB" will be displayed."

       This is how I ran snmpwalk - without giving a OID or name.

       So if there is some unrecognized device "IF" on the target system most likely an error message is displayed. Shouldn't caused the crash, right?

       Any pointing to the direction of searching the cause is highly appreciated. Thanks in advance.

-Jim
        [EMAIL PROTECTED]  P1??z?韔諲7?f?-８?+溣p)名????AⅣ那賃v屹{h?h?穖*.?????隩<+ay楓?h?+y羕 y呈???跱!z楚嗓?蔰蝍犯(?炮?5螔?n梓?言?%?Mz??津??l?叛賥?囫趵y?梁?賥?0?)j迫濇劂?煍?jxＪ^?m??菈??玒?叛賥?囫蜌?l?炮)舝╦傾璁?