BGP counter graph has a lot of holes

[rony]

Unless you deleted the file manually, it is still there, you just have to go find it..

[Wunk]

Ah right.., I sent them by E-mail, I think those are the ones considering they weren't updated since this morning when I deleted the cacti items..

[egironda]

I had this exact issue, with this exact template. I've never found the answer. Maybe I'll try recreating them and see if it helps

[xbertz]

I have the same issue happening. I am running the following:

Cacti Version - 0.8.6i
Plugin Architecture - 1.0
Poller Type - cactid.php
Server Info - Linux 2.6.9-42.0.3.ELsmp
Web Server - Apache/2.0.52 (CentOS)
PHP - 4.3.9
MySQL - 4.1.20
RRDTool - 1.2.15
SNMP - 5.1.2

I am noticing the holes only in the 5 min ave graphs (daily.) The weekly and other graphs are fine. Here is the output of the command recommended:
[root@netmon html]# cactid/cactid --verbosity=5 9 9
CACTID:

It also doesn't seem to matter whether or not I am using cactid or cmd.php. Any help would be great! Thanks in advance.
Here is a sample of th 5 min graph:

[Wunk]

I solved it by deleting de graph+datasource and recreating them.

[scardus]

I am seeing exactly the same problem on my Cacti server...

According to my logs, I can see cacti polling my routers and getting results back every 5 minutes. Here's a log extract from one of my BGP peers where a gap appears on my graph (hostname/ip addresses removed):

01/16/2007 11:55:20 PM - CMDPHP: Poller[0] Host[9] DS[1215] SNMP: v2: router.hostname, dsname: bgpPeerOutTotalMsg, oid: .1.3.6.1.2.1.15.3.1.13.111.222.333.444, output: 74854

01/16/2007 11:55:20 PM - CMDPHP: Poller[0] Host[9] DS[1215] SNMP: v2: router.hostname, dsname: bgpPeerOutUpdates, oid: .1.3.6.1.2.1.15.3.1.11.111.222.333.444, output: 27

01/16/2007 11:55:20 PM - CMDPHP: Poller[0] Host[9] DS[1215] SNMP: v2: router.hostname, dsname: bgpPeerInUpdates, oid: .1.3.6.1.2.1.15.3.1.10.111.222.333.444, output: 3359524

01/16/2007 11:55:20 PM - CMDPHP: Poller[0] Host[9] DS[1215] SNMP: v2: router.hostname, dsname: bgpPeerInTotalMsg, oid: .1.3.6.1.2.1.15.3.1.12.111.222.333.444, output: 3434250

I've tried deleting the graphs & data-sources & re-creating without any success. Every other type of graph on my server works fine...

If it makes any difference, I'm using 32-bit binaries of RRDTool on a 64-bit Centos v4.4 installation - The reason for this is so I can easily mirror the RRD files to other servers which may or may not be 64/32-bit arch's.

Any ideas?

[scardus]

I'm still trying to debug the gaps I'm seeing in my Cacti BGP graphs...

I've enabled debug logging, and can see RRDTool update being called with the values gleaned from SNMP and no errors being generated.

I've attached another graph showing a gap - It appears that we've got a value for 13:10 & 13:15, but for some reason no line is being drawn between the two. Can anyone shed a light on why this might be happening?

TIA

[scardus]

I've attached another graph showing a gap - It appears that we've got a value for 13:10 & 13:15, but for some reason no line is being drawn between the two. Can anyone shed a light on why this might be happening?

Hmm - I think I might have found my problem - There's NaN's in my RRD file at the timestamps where a gap appears:

Code: Select all

<xport>
  <meta>
    <start>1170680400</start>
    <step>300</step>
    <end>1170681900</end>
    <rows>6</rows>
    <columns>1</columns>
    <legend>
      <entry>Export</entry>
    </legend>
  </meta>
  <data>
    <row><t>1170680400</t><v>4.2406648086e-01</v></row>
    <row><t>1170680700</t><v>NaN</v></row>
    <row><t>1170681000</t><v>3.2214765101e-01</v></row>
    <row><t>1170681300</t><v>NaN</v></row>
    <row><t>1170681600</t><v>5.7525083612e-01</v></row>
    <row><t>1170681900</t><v>5.8029022918e-01</v></row>
  </data>
</xport>

The poller log entry that corresponds to the first NaN is:

Code: Select all

02/05/2007 01:05:23 PM - POLLER: Poller[0] CACTI2RRD: /usr/bin/rrdtool update /var/www/html/cacti/rra/router_bgppeerouttotalmsg_1215.rrd --template bgpPeerOutTotalMsg:bgpPeerOutUpdates:bgpPeerInUpdates:bgpPeerInTotalMsg 1170680716:15649:2:530395:546042

I'm not seeing any errors in my cacti logs with debug turned on, can anyone suggest a next step?

TIA

[gandalf]

You surely found my "NaN Debugging Help", linked in my signature? I suspect a given MAX value that is exceeded on a single poll
Reinhard

[scardus]

Hi,

Thanks for the reply...

You surely found my "NaN Debugging Help", linked in my signature? I suspect a given MAX value that is exceeded on a single poll

I found the document and followed all of the checks and I'm still stumped. In all my BGP RRDs, the max value is set to NaN which is meant to be "unlimited"...

Code: Select all

rrd_version = "0003"
step = 300
last_update = 1170695716
ds[bgpPeerOutTotalMsg].type = "COUNTER"
ds[bgpPeerOutTotalMsg].minimal_heartbeat = 300
ds[bgpPeerOutTotalMsg].min = 0.0000000000e+00
ds[bgpPeerOutTotalMsg].max = NaN
ds[bgpPeerOutTotalMsg].last_ds = "15899"
ds[bgpPeerOutTotalMsg].value = 2.6666666667e-01
ds[bgpPeerOutTotalMsg].unknown_sec = 0
ds[bgpPeerOutUpdates].type = "COUNTER"
ds[bgpPeerOutUpdates].minimal_heartbeat = 300
ds[bgpPeerOutUpdates].min = 0.0000000000e+00
ds[bgpPeerOutUpdates].max = NaN
ds[bgpPeerOutUpdates].last_ds = "2"
ds[bgpPeerOutUpdates].value = 0.0000000000e+00
ds[bgpPeerOutUpdates].unknown_sec = 0
ds[bgpPeerInUpdates].type = "COUNTER"
ds[bgpPeerInUpdates].minimal_heartbeat = 300
ds[bgpPeerInUpdates].min = 0.0000000000e+00
ds[bgpPeerInUpdates].max = NaN
ds[bgpPeerInUpdates].last_ds = "541421"
ds[bgpPeerInUpdates].value = 2.1440000000e+01
ds[bgpPeerInUpdates].unknown_sec = 0
ds[bgpPeerInTotalMsg].type = "COUNTER"
ds[bgpPeerInTotalMsg].minimal_heartbeat = 300
ds[bgpPeerInTotalMsg].min = 0.0000000000e+00
ds[bgpPeerInTotalMsg].max = NaN
ds[bgpPeerInTotalMsg].last_ds = "557318"
ds[bgpPeerInTotalMsg].value = 2.1706666667e+01
ds[bgpPeerInTotalMsg].unknown_sec = 0

[kplau]

urg.. I'm starting to have the very same issue you guys were having.. holes in graph. are there any resolution yet to this problem ? would be a shame not being able to use this great bgp graph template...

thanks in advance!

[gandalf]

Hi,

Thanks for the reply...

You surely found my "NaN Debugging Help", linked in my signature? I suspect a given MAX value that is exceeded on a single poll

I found the document and followed all of the checks and I'm still stumped. In all my BGP RRDs, the max value is set to NaN which is meant to be "unlimited"...

Yep, I agree. It is set to "unlimited". What are your results on the other steps mentioned in my HowTo?
Reinhard

[scardus]

Yep, I agree. It is set to "unlimited". What are your results on the other steps mentioned in my HowTo?

It's been a while since I last looked into this, but your reply made me go back and investigate once more.

I've just gone over the entire config again, and I think I might have found the problem.

I went over my logs & rrd files with a fine toothed comb and found that even though an rrdtool update was being called with correct values, occasionally just a NaN would be entered.

The BGP XML templates I have here specify a step & heartbeat of 300 seconds - I suspect what's happening is occasionally an update isnt happening in time and is being ignored by rrdtool.

I've rrdtune'd one of my rrd files with a heartbeat of 600 and will post my results shortly...

Sean

[scardus]

I've rrdtune'd one of my rrd files with a heartbeat of 600 and will post my results shortly...

Looks like it did the job...

[kplau]

took me awhile to figure out how to change the heartbeat.. heh.. just look in the xml file and changed all those 300 to 600..and like you said.. works like a charm!

Thanks scardus!