/ports/ repeats ports about librenms HOT 12 CLOSED

I've checked this and mine doesn't appear to be doing this. Anyone else having issues with this?

from librenms.

I'm not seeing this either, on any of my Cisco, Juniper, Mikrotik, or Ubiquiti devices. Could it be a device-specific bug?

from librenms.

I'm monitoring 5 Debian servers and after some scrolling all ports appear double in the same order as before.

I will try to boot up the vm containing librenms and make a screenshot for before catching my flight.

On 17.12.2013, at 18:22, Paul Gear wrote:

I'm not seeing this either, on any of my Cisco, Juniper, Mikrotik, or Ubiquiti devices. Could it be a device-specific bug?

—
Reply to this email directly or view it on GitHub.

from librenms.

I'm not sure I follow still, are you saying if you scroll (browser scroll) up and down on the /ports/ page it generates duplicates?

from librenms.

Hi,

sorry for the late reply, family ate up all my resources with this xmas
junk...

Yes, whenever I scroll down, I see the device's ports being printed
again.

Is this more of a feature than a bug? If so, I'm sorry for the false
alarm.

Happy New Year btw!

f0o

Security Research and Voodoo fanatic

On 2013-12-21 21:54, laf wrote:

I'm not sure I follow still, are you saying if you scroll (browser scroll) up and down on the /ports/ page it generates duplicates?

Reply to this email directly or view it on GitHub [1].

Links:

[1] #30 (comment)

from librenms.

I know the feeling with Xmas :)

Can you provide anymore details, OS you are using, browser version and number? What devices and number are you monitoring with how many ports?

I still can't replicate this :(

Happy new year to you as well :)

from librenms.

@f0o As suggested by @laf, please provide the OS, browser, and browser version you're using when you see this issue. If you can provide external guest-level access, this would help us even more as we could see for ourselves and use client-side debugging tools.

Assuming that's not possible, could you please provide screenshots and also run Chrome's Developer Tools (if you're using Chrome) or Firebug (if Firefox) and see if maybe there's a weird AJAX call happening?

Without a refresh or AJAX call, new ports shouldn't magically show up.

from librenms.

Hi,
sorry for the late reply again... my mailserver's HDD crashed..

I was using Opera 12.16 on a Linux.

The Hosts monitored were stock debians and gentoos, just a test-setup with 3-5 systems.
LibreNMS was installed on a fully patched debian wheezy using.

I cant provide access because the box was just an internal test.
However I will boot up the VM on the weekend (or earlier, depends on how much work I get done) and provide screenshots.
I can check for nifty AJAX calls too, maybe it's opera's JS-Engine that's looping over a json twice..

Cheers,
f0o

from librenms.

Hi,
the bug was related to Opera's JS-Engine.
It doesnt exist in the up2date version of Opera anymore.

I guess this solved itself then, sorry for wasting your attention.

Have a nice week!

from librenms.

So this issue just occurred again.

I figured out why, it's an issue with the Memcached check within html/pages/ports.inc.php

Working on a fix.

//Edit:
does anybody know if the memcached prefixes etc changed?
A restart of memcached to clear out the caches cleared the issue (obviously), now let's wait until the cache gets propagated again....

//Edit2:
Ok the poller populates the index 'ports-$ID-state' and the html-page merges that into the SQL results.
This obviously causes duplications everywhere...
Gonna work on a fix now

from librenms.

Not sure if it's related to this, but I also get duplicate BGP/iBGP and OSPF entries under routing.

from librenms.

The best fix for this particular issue is to remove the memcached checks on ports.inc.php. It has no use whatsoever. It was meant to relief the MySQL from pulling so much data but it's healthier for the SQL to handle one big * query than potentially hundreds * queries with 1 result each. So the Memcache pulling that's done afterwards will always just create duplicates. I will make a PR to remove those 3-4 lines.

@gadago it seems to be an unrelated similarity, the routing display makes no use of Memcached so it's duplications have a different cause...

from librenms.