From: xbgmsharp <xbg...@gm...> - 2007-12-19 15:18:58
|
Hi all, After a day of DEBUG. I works. I modify the file dm_snmp.pm which do snmp pooling. So now instead of pooling all leaf oid in one request, 1 oid equal 1 request. So now than i am pooling each oid separitly devmon should not go purple. I try to make it per pakets like 5 or 10 but devmon get lost when corresponding data with test done in dm_tests.pm I will let it working on some PIX and F5 devices and commit by the end of the week. All of this, is for not getting SNMP Error: "error status: tooBig". Because of this error all my leaf test was going clear. Comments? Regards, Francois xbgmsharp a écrit : > From this: > http://www.cisco.com/en/US/docs/net_mgmt/ciscoworks_ciscoview/4.1/quick/guide/cvgstrcv.html > > > SNMP Error Messages > tooBig: > The request you made cannot fit into a single packet. Generally, > CiscoView splits requests for physical view status until the device can > respond. In certain cases, CiscoView assumes that if an agent times out > on 20 or more variables, the agent might not be able to respond because > the request is too big; it splits the request and resends it. Check that > the MTU size on the SNMP interface is as large as possible so that > CiscoView does not waste bandwidth by sending more than one request. > > > > xbgmsharp a écrit : > >> Re, >> >> Could you add this line: >> >> my ($bindings) = $session->decode_get_response($response); >> if(!defined $bindings or $bindings eq '') { >> my $snmp_err; >> do_log("DEBUG SNMP MSG: $SNMP_Session::errmsg()",0) if $g{'debug'} >> ($snmp_err = $SNMP_Session::errmsg) =~ s/\n.*//s; >> >> on line 474 in the file modules/dm_snmp.pm. >> >> SNMP Error: >> Received SNMP response with error code >> error status: tooBig >> index 0 >> SNMPv2c_Session (....... >> >> I got this error the error code "error status: tooBig" is explain. >> http://www.juniper.net/security/auto/vulnerabilities/vuln2591.html >> >> it seen very weird. >> But if i remove a test everything is fine. Also if i pool it by hand >> with smnpwalk or snmpget it is working. >> Could you also test and tell me you have the same error msg. >> >> Here is a shell command which allow you test all oid from a template. >> >> find . -type f -name "oids" | xargs cat | cut -d ':' -f 2 | tr -d ' ' | >> grep -v "^$" | awk '{ print "snmpwalk -t 5 -On -v2c -c COMMUNAUTY IP " >> $1 }' | sh >> >> >> >> Francois. >> >> >> >> >> Nathan Hand a écrit : >> >> >>> On Tue, 18 Dec 2007 23:05:48 +0100, "xbgmsharp" <xbg...@gm...> >>> said: >>> >>> >>> >>>>> Short summary: If any single oid fails to retrieve then devmon stops >>>>> collecting the rest of the oids in the template. >>>>> >>>>> I've confirmed this hypothesis with with -f -p -vvvvvvvvvvv. Devmon >>>>> says something about "too many failures" and "skipping device". >>>>> Perhaps devmon should only skip that single column and still test the >>>>> others. >>>>> >>>>> >>>>> >>>> In the code, when 2 tests fails it stop chec the device and send a >>>> clear status. >>>> >>>> in modules/dm_snmp.pm >>>> >>>> 535 # We dont want to do every table if we are failing alot of >>>> walks >>>> 536 if($failed_query > 2) { >>>> 537 my $error_str = >>>> 538 "Failed too many queries on $dev, aborting query"; >>>> 539 $data_out{'error'}{$error_str} = 1; >>>> 540 send_data($sock, \%data_out); >>>> 541 $session->close(); >>>> 542 next DEVICE; >>>> 543 } >>>> >>>> >>>> I have the same problem with memory on my pix. >>>> >>>> But it never made the test go purple. >>>> >>>> >>>> >>> The interesting thing is it makes _other_ tests go purple. >>> >>> My guess is that because it aborts the device query it doesn't send any >>> message for subsequent tests on that device, not even clear messages. So >>> after an hour the other tests go purple. >>> >>> >>> |