Okay per the grep below, the problem is in memory_netbsd1.c,
which, given the name of the file, would explain a great deal of
desynchronization between the kernel data structures and the state of the
MIB data retrieval algorithm. The logic is the line:
if (swapFree < minimumswap)
....blah
In FreeBSD, swapFree get's assigned
swapFree += kswap[i].ksw_total
But in OpenBSD (I'm assuming it uses the NetBSD #ifdef'd file), I don't
really even see where swapFree is getting set. Unless
long_ret = (swapFree > minimumswap) ? 0 : 1;
...bitwise operation is doing so. But I doubt it. I suspect the problem
is persistent in NetBSD as well (CC'ing pkgsrc maintainer). Also, I don't
know how far apart the NetBSD and OpenBSD virtual memory systems have
grown.
net-snmp-5.2.1.2/agent/mibgroup/ucd-snmp$ grep -ir swapFree ./*
./memory.c: unsigned long *swapfree);
./memory.c: swapfree;
./memory.c: &swapfree);
./memory.c: unsigned long *swapfree)
./memory.c: b = strstr(buff, "SwapFree: ");
./memory.c: sscanf(b, "SwapFree: %lu", swapfree);
./memory.c: snmp_log(LOG_ERR, "No SwapFree line in
/proc/meminfo\n");
./memory.c: *swapfree = 0;
./memory.c: swapfree
./memory.c: case swapfree:
./memory.c: spaceleft = memory(swapfree);
./memory.c: long_ret = memory(memfree) + memory(swapfree);
./memory_freebsd2.c:quad_t swapFree;
./memory_freebsd2.c: sscanf(ext.output, "%*s%*d%qd%qd", &swapUsed,
&swapFree);
./memory_freebsd2.c: swapTotal = swapUsed + swapFree;
./memory_freebsd2.c: swapUsed = swapTotal = swapFree = 0;
./memory_freebsd2.c: swapFree += kswap[i].ksw_total -
kswap[i].ksw_used;
./memory_freebsd2.c: swapUsed = swapTotal - swapFree;
./memory_freebsd2.c: swapFree *= pagesize;
./memory_freebsd2.c: long_ret = swapFree;
./memory_freebsd2.c: long_ret = (swapFree > minimumswap) ? 0 : 1;
./memory_freebsd2.c: if (swapFree < minimumswap)
./memory_freebsd2.c: sprintf(errmsg, "Running out of swap space
(%qd)", swapFree);
./memory_netbsd1.c:quad_t swapFree;
./memory_netbsd1.c: long_ret = (swapFree > minimumswap) ? 0 : 1;
./memory_netbsd1.c: if (swapFree < minimumswap)
./memory_netbsd1.c: sprintf(errmsg, "Running out of swap space
(%qd)", swapFree);
On Mon, 15 Aug 2005, Brian A. Seklecki wrote:
[with the correct address list this time]
All:
Another issue with OpenBSD. Using the 'defaultMonitor' config with OpenBSD
3.7/i386 running Net-SNMP 5.2.1p1. At agent restart, I always get a trap
from the agent about swap:
example:
Jul 25 08:49:01 nmsHostname snmptrapd[72648]: 2005-07-25 08:49:01
hostname.com [UDP: [192.168.1.1]:45160]: DISMAN-EVENT-MIB::sysUpTimeInstance
= Timeticks: (40857874) 4 days, 17:29:38.74 SNMPv2-MIB::snmpTrapOID.0 = OID:
DISMAN-EVENT-MIB::mteTriggerFired DISMAN-EVENT-MIB::mteHotTrigger = STRING:
memory DISMAN-EVENT-MIB::mteHotTargetName = STRING:
DISMAN-EVENT-MIB::mteHotContextName = STRING: DISMAN-EVENT-MIB::mteHotOID =
OID: UCD-SNMP-MIB::memSwapError.0 DISMAN-EVENT-MIB::mteHotValue = INTEGER: 1
UCD-SNMP-MIB::memErrorName.0 = STRING: swap UCD-SNMP-MIB::memSwapErrorMsg.0 =
STRING: Running out of swap space (0)
I'm not sure where DISMAN-EVENT-MIB is pulling information about swap
(sysctl? kernel? userland?), but it's erroneous in this case.
# swapctl -l
Device 1K-blocks Used Avail Capacity Priority
/dev/raid0a 2097024 0 2097024 0% 0
In UCD-SNMP-MIB.txt, the following is defined:
memMinimumSwap OBJECT-TYPE
SYNTAX Integer32
MAX-ACCESS read-only
STATUS current
DESCRIPTION
"Minimum amount of free swap required to be free
or else memErrorSwap is set to 1 and an error string is
returned memSwapErrorMsg."
::= { memory 12 }
And that traces to net-snmp-5.2.1.2/agent/mibgroup/ucd-snmp/memory.h where
#define MEMSWAPMINIMUM 12
...I'll look into how FreeBSD and OpenBSD accumulate this value differently.
It's burried in the source somewhere.
l8*
-lava
x.25 - minix - bitnet - plan9 - 110 bps - ASR 33 - base8
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
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
l8*
-lava
x.25 - minix - bitnet - plan9 - 110 bps - ASR 33 - base8
-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf
_______________________________________________
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