[SOLVED]rrdtool and rrd file question

[eholz1]

Hello Members,
I am running cact1-1.1.28, I have a cisco router.
I am monitoring 4 interfaces. One of the 4 creates a graph with
a gap periodically.
Is there a way I can use rrdtool info or other rrdtool feature to find how/where the
gap is being created? cron runs the poller every 5 min, and spine runs every
30 seconds.

I can change the graph, to show the gap based on the graph start and end time.
The rrd file and graph are being updated correctly. Also, strange, the poller is set for
update every 30sec, but the datasource shows it updating every 5 minutes, aka 300s step.
how does that happen?

Any Ideas on this?

eholz1

SOLVED - more or less - chicken or the egg, etc.

[netniV]

That happens if you change the profile after already beginning to graph. When RRDTool first creates its database file, it puts the steps/interval etc into it then from what I recall.

[eholz1]

hello netniv,

Hmmm. When you say profile, I will assume you mean the "Data Profiles" under cacti Presets.
I have inherited a cacti installation which has two profiles defined using the same name.
This install also has some rrd files with datasource info like this:
Number of rows for Cacti RRA id '0' should be '600'
RRA 'cacti_1' missing in Cacti definition
RRA 'cacti_2' missing in Cacti definition
RRA 'cacti_3' missing in Cacti definition
Number of rows for Cacti RRA id '4' should be '600'
RRA 'cacti_5' missing in Cacti definition
RRA 'cacti_6' missing in Cacti definition
RRA 'cacti_7' missing in Cacti definition

BUT: thee rrd files continue to be updated, and graphs are created OK. I can fix the issue of those "missing" definitions by
enabling the datasource debug mode, copying the file update code, and recreating the file - this creates an rrd and datasource with the
"correct" number of definitions (maybe). I have no idea how the extra missing defintions got there in the first place - but there are
many such files with that issue. Too many to run a script to correct the issue. This cacti system started with 0.8, and was upgraded over
time to 1.1.28. I plan to attempt another upgrade to 1.1.38.

Having two profiles with the same name does not make any sense to me. What do you think?

So I did not delete one of the profiles, I just renamed one to Upgrade Profile 2. So I still have
two data profiles, Upgrade Profile 1 and Upgrade Profile 2. I have a total of 3770 datasources.
3480 with Profile 1, and 290 with Profile 2. Profile 1 has a heartbeat of 1min, and Profile 2 has a
heartbeat of 2min.

Another interesting thing - Profile 2 has 290 datasources, and each device's datasource is for the Cisco router/switch
"Cisco Router - 5 Minute CPU" - I am thinking we do not need this at all!

After this renaming, it looks like my system is graphing/updating better. Is it possible that when the data
profiles had the same name - some sort of conflict was arising?

I may be further along in solving this mystery.

Thanks for your valuable input.

Eholz1