Nagios Support Forum

Upgrade to 2014R1.0 blew away my MRTG file. Is this something new?

Boy, this was not a good idea!!!


CentOS 5.7 x64
physical box using the xi-xxxxx.tar.gz

My graphs seem to be working. Maybe I missed a note or something.

Added note. While your upgrade scripts migrate mrtg, from what i can tell, it left about half or more of my hosts without migration.

The mrtg config should no longer store anything other than the directory for the other configurations as individual devices. Could you provide an example of one that failed please? To my knowledge there shouldn't be any issues with the migration script in particular, but there is always a chance with different devices and configurations.

It looks like you now use a conf.d directory to store the individual files, for each device. Is that true?

However, my main switch was not migrated at all, IOW, there was no x.x.x.x.cfg created for that device in conf.d

I am also noticing all my graphs do not seem to be indicating any traffic, and even my ping RTA graphs for all devices are now indicating nothing. Yet I am not getting alerts. So I guess the graphing piece has gone south on this upgrade

You are correct that they are now in the conf.d directory. Have you tried grepping the configs, it is entirely possible that it was merged with another config, although slightly strange that it would do that.
Also I do believe that you should have an mrtg.confg.bk in /etc/mrtg/ still, is that true and does it have your device listed in there?

I did, and no they did not show up in any other file. I manually transferred the data. I'm waiting to see... but as I mentioned, my ping charts and everything are not displaying now either.

Charting is just not working at all now.

Oh, your upgrade did place a bunch of entries in another single config.

Just an FYI, you guys may wish to look at what your update script is keying off of. Over the years I have had to make various manual entries in my mrtg.cfg file, even with your assistance and input. Well, those entries did not always have headers, etc in place. I believe this is where your update script fails. I have had several device cfg files that contained multiple IP addressing configurations for my devices, after upgrading. While, technically it works, it makes spaghetti out of the process, and now I'm combing thru every single device config looking for other devices configurations.

Keeping a small, error-free mrtg.cfg file, and placing multiple configs in the "/etc/mrtg/conf.d/" directory is a good thing in the majority of the cases. It would be easier to rerun the wizard, in cases when you want to add ports (or some ports were administratively down during the first scan). If the past, you needed to remove all of the "old" entries from the "mrtg.cfg" file, which could be a tedious process. Having separate files, and not having to dig through one huge file could save you some valuable time. It would also be easier for us (on the support side of things).

Hello,

I have this issue resolved. I understand why you perform the migration, but I think the method needs review.

I also wanted to let you know that when you performed the migration, that you did not backup the original mrtg.cfg (yes, technically it was up to me, and I did). You created an mrtg.cfg.orig, but it was an empty default. My thought was that you could have easily used that file to store the customer's original, before blowing it away.

-j