rrdtool graphs/Reporting broken image links

[Fraoch]

Hello:

This just started today - some of the graphs created by rrdtool through the "Reporting" function show as broken after a few minutes. See the attachment.

I can temporarily fix this by zooming out on the nonexistent graph and zooming back in again, but in a minute or two the links get broken again.

Is there anything I can check?

If it's browser related, I'm using Firefox 33.0 on Linux Mint 17.

 

[cyberjock]

I don't have a fix, but I asked a dev about this. If we can find a way to reproduce this we should be able to identify and fix the bug.

 

[Fraoch]

Thanks, but this all could be a red herring as the web interface is currently slowly failing the same way I saw here:

https://forums.freenas.org/index.php?threads/zio-io_cv-messages-on-boot.24306/

I'm doing an intense iozone test now and it's slowly becoming unresponsive, starting with the graphs. I thought it might be the browser and cleared the cache...now my left panel is gone (see attachment). I also notice my scrolling messages on the bottom are missing.

It's not totally unresponsive but it's failing slowly. I still have SSH access and there still seems to be plenty of CPU and memory resources to spare so I'm not sure what's going on here. Although the test is intensive, the disks actually aren't all that busy, 355.1 kB/s read currently.

So it might not be applicable to FreeNAS as a whole, but just my system.

 

[cyberjock]

Were you doing an iozone test when you had the problem that you mentioned in post #1?

iozone is a grueling test on ZFS. It's hardcore and mostly unrealistic as it (when properly setup for appropriate settings) finds the limits of your hardware. This obviously isn't how the pool should be allowed to operate in real-world, so we can expect problems with testing the pool with theoretically unexpected workloads. But interesting nonetheless. ;)

 

[Fraoch]

Were you doing an iozone test when you had the problem that you mentioned in post #1?

Yes.

iozone is a grueling test on ZFS. It's hardcore and mostly unrealistic as it (when properly setup for appropriate settings) finds the limits of your hardware. This obviously isn't how the pool should be allowed to operate in real-world, so we can expect problems with testing the pool with theoretically unexpected workloads. But interesting nonetheless. ;)

Yeah I'm mostly doing it as a stress test on the drives and the system and indeed, it looks like it's stressed.:p It's running the most stressful type of test I could find as this situation is occurring. It doesn't look like it's particularly stressed at the moment - CPU is mostly idle, some free memory and drives only reading in the ~300 kB/s range (kilobyte) so I'm surprised that the web interface is failing but if that's the worst that can happen I'm good. The data hasn't been corrupted and the drives don't have any errors.