Differing rrd_id graphs post upgrade from 0.8.8h to 1.1.37

[frekemeister]

Hi Guys,

Finally got round to upgrading from an aging first gen Raspberry Pi running Cacti 0.8.8h to a Pi 3 and upgrading to version 1.1.37

Not quite sure how the cmd.php was ever scheduled in 0.8.8h as I struggled to find any reference of a cron job anywhere, which had me stumped for a few hours after the upgrade as I was seeing no poller activity. Creating the necessary user and crontab sorted this.

To my point - I use an external (to Cacti) website to display a various monitoring screens, since 0.8.8 the rrd_id time periods on the graphs seem to have changed, for instance:

0.8.8
rrd_id=0 - the last 24 hours
rrd_id=1 - the last 24 hours
rrd_id=2 - the last week
rrd_id=3 - the last month
rrd_id=4 - the last year
rrd_id=5 - the last 4 hours

1.1.37 (as I can best determine)
rrd_id=0 - the last 24 hours
rrd_id=1 - the last 48 hours
rrd_id=2 - the last 2 weeks
rrd_id=3 - the last 3 months
rrd_id=4 - the last 2 years
rrd_id=5 - the last 48 hours

How/where are these rrd_id details configured/can they be changed back to what I had before. Got some odd looking graphs on my monitoring screens?

Comparison attached - rrd_ids go from 0->5, 1.1.37 on the left 0.8.8h on the right.

Cheers guys!

[Osiris]

Post a screen shot of your two data source profiiles, in detail (edit screen), then the rrdtool info output from those two files.

[Osiris]

Second thing, in 1.1.37, go into the Data Source, and press the "Info Mode" and post that screen shot. It's like a RRDtool info command on steroids.

[frekemeister]

Hi Osiris,

Thank you for getting back to me, I've attached a screenshot of the 1.1.37 "info mode", however I'm struggling to find where the rrdtool info output for the detail can be found - I have attached the Data Source Debug just in case.

Karl

[Osiris]

Your issue is common for people who come from 0.8.8 or before. At some point in the old install, someone selected the 1 minute RRA and added it to your data template. As a result, you have a mess on your hands. Post the Data Source Profile edit page as I requested. You may have more than one.

[frekemeister]

Hi Osiris,

See attached for 1.1.37

[frekemeister]

RRA configuration in 0.8.8h

[Osiris]

Profile 1 is bad. From the main profile page, how many data sources and templates affected?

[frekemeister]

Going from my screenshots (as both profiles currently have the same name), DS Profile1.jpg there are 35 data sources, DS Profile2.jpg there are 239 data sources using the profile.

Which screenshot were you referring to as "Profile 1"? I guess it would help if I renamed the profiles..

[Osiris]

The main Data Source Profile screen. You should see a count there. The solution is a few fold and varies on if you want to preserve data.

1) Don't preserve data.
a) Change All Data Templates to the good Data Source Profile
b) Delete all Graphs that used the bad Data Source Profile
c) Recreate them

2) Want to preserve data
a) Change All Data Templates to the good Data Source Profile
b) Update the data_source_profile_id in the table data_template_data to the good Data Source Profile
c) Rename all the RRDfiles that have bad data to a backup name
d) Copy the RRDfile create syntax into a script
e) For each RRDfile impacted, perform successive fetches 5 minutes apart from two years ago to now, inserting into a temporary RRDfile that you just created using that syntax above. You will have to learn how to insert into RRDfiles using RRDtool.
f) Rename the file back to the one Cacti expects.

I can not write the script for you. But you might be able to find someone who can.

[frekemeister]

Hi Osiris,

Thank you for the information, appreciated. I'd ideally like to retain the data, am I right in assuming at step f) I need to rename the old RRD file back to the original name?

[netniV]

No, it would be the temporary file that you inserted data into using RRDTool

[Osiris]

And then, in the end, renamed back to the original rrdfile. Make sure permissions are right afterward too.

[frekemeister]

Awesome, thanks both.

As a side note, is there harm being done leaving the datasources as is for now?