mirror of
https://github.com/munin-monitoring/contrib.git
synced 2025-07-21 18:41:03 +00:00
[snmp__if_combined] Add support for stacked series in root graph
From a summary comment in the code:
If a stacked graph is requested, plot the total. Rather than
aggregating them with CDEFs of SUMs, we sum them in the script, which
allows us to create real series than can be easily borrowed by other
loaning graphs.
Those series are recv_bits and send_bits.
There are longstanding bugs with them (dating back to snmp__if). This
has been documented, as well as some potential ad hoc fixes for the
data.
Some outdated bug were also removed from the doc.
Signed-off-by: Olivier Mehani <shtrom@ssji.net>
This commit is contained in:
Olivier Mehani
parent
61473dfdb0
commit
6ac458ace1
3 changed files with 151 additions and 35 deletions
BIN
plugins/snmp/example-graphs/snmp_if_combined-1.png
Normal file
BIN
plugins/snmp/example-graphs/snmp_if_combined-1.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 53 KiB |
BIN
plugins/snmp/example-graphs/snmp_if_combined-day.png
Normal file
BIN
plugins/snmp/example-graphs/snmp_if_combined-day.png
Normal file
Binary file not shown.
|
After Width: | Height: | Size: 70 KiB |
|
|
@ -13,6 +13,24 @@ as "munin-node-configure --snmp switch.langfeldt.net --snmpversion 2c
|
|||
interfaces. On a typical switch you will get one plugin pr. ethernet
|
||||
port. On a router you might get one plugin per VLAN interface.
|
||||
|
||||
=head2 NOTABLE DIFFERENCE WITH SNMP__IF_MULTI
|
||||
|
||||
The smnp__if_multi plugin records data in _bytes_ per second, and shows
|
||||
them as bps by using CDEFs when graphing. This plugin performs the
|
||||
multiplication upon reporting the value, and therefore doesn't have a
|
||||
CDEF in the graphing path.
|
||||
|
||||
This is important when migration from snmp__if_multi, for at least two aspects:
|
||||
|
||||
- Reusing RRDs: it is not possible to just rename the old RRD files from
|
||||
smnp__if_multi, as historical data will be graphed as 8 times smaller than in
|
||||
reality. The stored values would need to be adjusted.
|
||||
- Data loaning: borrowing graphs no longer need to include a CDEF to convert
|
||||
bytes to bits. Conversely, they do now need a CDEF to convert bits to bytes.
|
||||
|
||||
To reflect this aspect explicitely, the root graph's totals are named
|
||||
recv_bits and send_bits.
|
||||
|
||||
=head1 CONFIGURATION
|
||||
|
||||
As a rule SNMP plugins need site specific configuration. The default
|
||||
|
|
@ -22,6 +40,7 @@ configuration (shown here) will only work on insecure sites/devices:
|
|||
env.version 2
|
||||
env.community public
|
||||
env.ifTypeOnly ethernetCsmacd
|
||||
env.stackedRoot 1
|
||||
|
||||
In general SNMP is not very secure at all unless you use SNMP version
|
||||
3 which supports authentication and privacy (encryption). But in any
|
||||
|
|
@ -51,12 +70,21 @@ would normally be for VPNs. A minor horde of different interface types
|
|||
are supposted, please see IANAifType-MIB (on your system or find with
|
||||
Google) for a full list.
|
||||
|
||||
The stackedRoot option determines whether the root summary graph shows the traffic
|
||||
on each interface separately, or stacks them on top of one another to show the
|
||||
total traffic through the device.
|
||||
|
||||
=head1 INTERPRETATION
|
||||
|
||||
A single graph is generated with all the interfaces overlaid one over
|
||||
the other: incoming traffic is received on the interface from the
|
||||
connected device, outgoing is sent to it instead.
|
||||
|
||||
With the `stackedRoot` option, traffic is plotted as stacked areas instead, and
|
||||
two additional series are calculated to show the total usage of the device.
|
||||
Those series are useful for loaning into higher-level summary graphs (but see
|
||||
BUGS below).
|
||||
|
||||
Sub-graphs are created, one per interface, akin to snmp__if_multi
|
||||
plugin.
|
||||
|
||||
|
|
@ -82,6 +110,8 @@ counters (see FEATURES below).
|
|||
|
||||
=head1 BUGS
|
||||
|
||||
=head2 Missing indexing options
|
||||
|
||||
Should support indexing by
|
||||
|
||||
- ifIndex
|
||||
|
|
@ -92,12 +122,41 @@ Should support indexing by
|
|||
|
||||
(and anything else MRTG can index by) in addition to OID-index as now.
|
||||
|
||||
Pulling in a user definable set of ifName/ifDescr/ifAlias for textual
|
||||
description and even graph_title would also be nice.
|
||||
=head2 Total traffic oddities in stackedRoot mode.
|
||||
|
||||
When using the `stackedRoot` option, the total lines (`send_bits` and
|
||||
`recv_bits`) may sometimes jump to nonsensical values. This is a bug shared with
|
||||
the original snmp__if_multi.
|
||||
|
||||
This seems to be due to the way those series are calculated, by summing the
|
||||
current values of the other counters. The first issue is that of summing 64-bit
|
||||
values into a single variable that is likely to also be 64 bits. The second
|
||||
issue (compounded by the first) is that of wraparound. When one of the counters
|
||||
wraps around, the sum jumps backwards. With a `min` set to 0, and other counters
|
||||
having kept increasing, this looks like a huge increase the total counter.
|
||||
Depending on the `max` (which should match the backplane bandwidth), this may
|
||||
not be correctly recognised as a spurious value, just reconded as valid.
|
||||
|
||||
There is no clear solution to this bug at the moment, save for trying to salvage
|
||||
the data after the fact. This can be done my either clipping all values beyond a
|
||||
maximum (e.g., the known use of the switch, rather that its full backplane
|
||||
bandwith).
|
||||
|
||||
RANGE='[0-9.]\+e+\(0[789]\|[1-9][0-9]\)' # Anything >=1e07
|
||||
rrdtool dump ${RRD_FILE} \
|
||||
| sed "s/<v>${RANGE}/<v>NaN/" \
|
||||
| rrdtool restore - ${RRD_FILE}.new
|
||||
mv ${RRD_FILE}.new ${RRD_FILE} # Make sure you have a backup!
|
||||
|
||||
Alternatively, if you can determine the time at whiche the spurious value was
|
||||
recorded, you can just make that value a NaN.
|
||||
|
||||
TIMESTAMP="1623492000"
|
||||
rrdtool dump "${RRD_FILE}" \
|
||||
| sed "/${TIMESTAMP}/s/<v>.*<\/v>/<v>NaN<\/v>/" \
|
||||
| rrdtool restore - "${RRD_FILE}.new"
|
||||
mv ${RRD_FILE}.new ${RRD_FILE} # Make sure you have a backup!
|
||||
|
||||
If we get a patch to support the .oldname attribute then we may use
|
||||
that to let the operator change the indexing dynamically without data
|
||||
loss.
|
||||
|
||||
=head1 FEATURES
|
||||
|
||||
|
|
@ -117,7 +176,7 @@ usage of it, it is nothing the plugin can fix. In the future Munin
|
|||
may be able to run the plugin more often than the counter wraps
|
||||
around.
|
||||
|
||||
=head1 AUTHOR
|
||||
=head1 AUTHORS
|
||||
|
||||
Copyright (C) 2012 Diego Elio Pettenò.
|
||||
|
||||
|
|
@ -136,6 +195,8 @@ Reworked to snmp__if_multi by Nicolai Langfeldt.
|
|||
|
||||
Reworked to snmp__if_combined by Diego Elio Pettenò.
|
||||
|
||||
Stacked option by Olivier Mehani <shtrom+munin@ssji.net>
|
||||
|
||||
=head1 LICENSE
|
||||
|
||||
GPLv2
|
||||
|
|
@ -177,6 +238,11 @@ if (defined $ARGV[0] and $ARGV[0] eq "snmpconf") {
|
|||
exit 0;
|
||||
}
|
||||
|
||||
my $stackedRoot = 0;
|
||||
if (exists $ENV{'stackedRoot'}) {
|
||||
$stackedRoot = $ENV{'stackedRoot'};
|
||||
}
|
||||
|
||||
my $sysDescr = '1.3.6.1.2.1.1.1.0';
|
||||
my $sysLocation = '1.3.6.1.2.1.1.6.0';
|
||||
my $sysContact = '1.3.6.1.2.1.1.4.0';
|
||||
|
|
@ -550,11 +616,18 @@ graph_info This graph shows the total traffic for $host.
|
|||
END
|
||||
|
||||
print "graph_order";
|
||||
my @ifs;
|
||||
foreach my $if (sort {$a <=> $b} keys %{$snmpinfo}) {
|
||||
print " recv$if=snmp_if_combined.$if.recv send$if=snmp_if_combined.$if.send";
|
||||
push @ifs, "snmp_if_combined.$if";
|
||||
}
|
||||
if ($stackedRoot) {
|
||||
print " recv_bits send_bits";
|
||||
}
|
||||
print "\n";
|
||||
|
||||
# XXX: For some reason, Munin doesn't resolve AREASTACK properly in this case...
|
||||
my $area_or_stack = "AREA";
|
||||
foreach my $if (sort {$a <=> $b} keys %{$snmpinfo}) {
|
||||
my $alias = $snmpinfoX->{$if}->{ifAlias} || $snmpinfo->{$if}->{ifDescr} || "Interface $if";
|
||||
|
||||
|
|
@ -568,11 +641,40 @@ recv$if.label $alias
|
|||
recv$if.graph no
|
||||
send$if.label $alias
|
||||
send$if.negative recv$if
|
||||
END
|
||||
if ($stackedRoot) {
|
||||
print "send$if.draw $area_or_stack\n";
|
||||
$area_or_stack = "STACK";
|
||||
}
|
||||
}
|
||||
if ($stackedRoot) {
|
||||
|
||||
# If a stacked graph is requested, plot the total. Rather than
|
||||
# aggregating them with CDEFs of SUMs, we sum them in the script, which
|
||||
# allows us to create real series than can be easily borrowed by other
|
||||
# loaning graphs.
|
||||
#
|
||||
# Using CDEFs and SUMs proved difficult anyway. Due to bugs in Munin
|
||||
# 2.0's GraphOld core (negatives and multigraph), the graph_total
|
||||
# behaves weirdly: - only the positive total is plotted (but both
|
||||
# positive and negatives are calculated); - the graph_total is also
|
||||
# included in the single graphs, where it is clearly unecessary.
|
||||
|
||||
print <<END;
|
||||
recv_bits.label Total traffic
|
||||
recv_bits.type DERIVE
|
||||
recv_bits.min 0
|
||||
recv_bits.graph no
|
||||
send_bits.label Total traffic
|
||||
send_bits.type DERIVE
|
||||
send_bits.min 0
|
||||
send_bits.draw LINE1
|
||||
send_bits.colour 000000
|
||||
send_bits.negative recv_bits
|
||||
END
|
||||
}
|
||||
|
||||
print <<END;
|
||||
|
||||
multigraph snmp_if_combined_err
|
||||
graph_title All interfaces errors
|
||||
graph_args --base 1000
|
||||
|
|
@ -723,34 +825,18 @@ sub do_fetch_if {
|
|||
|
||||
my $status = $snmpinfo->{$if}->{ifOperStatus} || 1;
|
||||
|
||||
my $response;
|
||||
|
||||
# 1 means up
|
||||
# 2 means set to down
|
||||
# Everything else we ignore.
|
||||
|
||||
print "multigraph snmp_if_combined.$if\n";
|
||||
|
||||
if ($status == 2) {
|
||||
if ($status != 1) {
|
||||
# Interface is down
|
||||
print "recv.value U\n";
|
||||
print "send.value U\n";
|
||||
print "send.extinfo This interface is currently down.\n";
|
||||
goto if_errors;
|
||||
}
|
||||
|
||||
$response = $snmpinfoX->{$if}->{ifHCInOctets} || $snmpinfo->{$if}->{ifInOctets};
|
||||
printf("recv.value %s\n", defined($response) ? ($response * 8) : "U");
|
||||
|
||||
$response = $snmpinfoX->{$if}->{ifHCOutOctets} || $snmpinfo->{$if}->{ifOutOctets};
|
||||
printf("send.value %s\n", defined($response) ? ($response * 8) : "U");
|
||||
|
||||
if_errors:
|
||||
|
||||
print "multigraph snmp_if_combined_err.$if\n";
|
||||
|
||||
if ($status == 2) {
|
||||
print <<END;
|
||||
multigraph snmp_if_combined.$if
|
||||
recv.value U
|
||||
send.value U
|
||||
send.extinfo This interface is currently down.
|
||||
multigraph snmp_if_combined_err.$if
|
||||
errors_in.value U
|
||||
errors_out.value U
|
||||
discards_in.value U
|
||||
|
|
@ -759,9 +845,25 @@ total_in.value U
|
|||
total_out.value U
|
||||
errors_out.extinfo This interface is currently down
|
||||
END
|
||||
return;
|
||||
return (undef, undef);
|
||||
}
|
||||
|
||||
my $response;
|
||||
my $recv;
|
||||
my $send;
|
||||
|
||||
print "multigraph snmp_if_combined.$if\n";
|
||||
|
||||
$response = $snmpinfoX->{$if}->{ifHCInOctets} || $snmpinfo->{$if}->{ifInOctets};
|
||||
$recv = defined($response) ? ($response * 8) : undef;
|
||||
printf("recv.value %s\n", $recv || "U");
|
||||
|
||||
$response = $snmpinfoX->{$if}->{ifHCOutOctets} || $snmpinfo->{$if}->{ifOutOctets};
|
||||
$send = defined($response) ? ($response * 8) : undef;
|
||||
printf("send.value %s\n", $send || "U");
|
||||
|
||||
print "multigraph snmp_if_combined_err.$if\n";
|
||||
|
||||
my $errors = $snmpinfo->{$if}->{ifInErrors};
|
||||
my $discards = $snmpinfo->{$if}->{ifInDiscards};
|
||||
printf("errors_in.value %s\n".
|
||||
|
|
@ -781,6 +883,8 @@ END
|
|||
$discards || "U",
|
||||
($errors || 0) + ($discards || 0)
|
||||
);
|
||||
|
||||
return ($recv, $send);
|
||||
}
|
||||
|
||||
sub do_config {
|
||||
|
|
@ -806,6 +910,18 @@ if ($ARGV[0] and $ARGV[0] eq "config") {
|
|||
exit 0;
|
||||
}
|
||||
|
||||
my $recv = 0;
|
||||
my $send = 0,
|
||||
my $recv_if;
|
||||
my $send_if;
|
||||
|
||||
foreach my $if (sort {$a <=> $b} keys %{$snmpinfo}) {
|
||||
do_fetch_if($if);
|
||||
($recv_if, $send_if) = do_fetch_if($if);
|
||||
$recv += ($recv_if || 0);
|
||||
$send += ($send_if || 0);
|
||||
}
|
||||
if ($stackedRoot) {
|
||||
print "multigraph snmp_if_combined\n";
|
||||
print "recv_bits.value $recv\n";
|
||||
print "send_bits.value $send\n";
|
||||
}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue