# connections disparity?

[dsalch]

This seems a bit odd, so I wanted to report it to see if this is a potential problem or not. It certainly caused me some grief till I figured out what was happening.

I have had intermittent issues with new connections failing for a long time, many months, and I finally tracked it down. Curiously, it seems a "reporting issue" or just my own misunderstanding.

I run several servers inside the network that offer cloud services outside. (i.e bitcoin, litecoin, StorJ). There are moments when things get overloaded in # connections, but not bandwidth. During these moments, if the total # conenctions exceeded the router setting for "max connections" new connections would fail intermittently. That seems normal. However I never noticed this until now becasue I was looking in the wrong place.

Status -> Connected Hosts -> Hosts with Active Connections

This location lists the # connections active for each host, including gargoyle itself. Yet in those moments of overload, the total number of connections here did not exceed the setting for max connections.

Status -> Overview -> Device -> Connections

This location lists the total connections active and the max setting. Here the # connections did reach the max setting when things quit working.

For instance, the hosts section would list about 2000 total connections but on the overview section it would show 4095. This was during a time when new connections failed. I increased the max connection setting to 8192 and the connections on the overview screen jumped up to about 6000 very quickly. The hosts section added up to about 4000.

After running a while, the # connection in both screens dropped, I believe becasue the application that was causing hte connections was doing frequent retires upon failing that led to the constantly climbing numbers. Once the max setting was increased, the active number rarely exceeds the 2000 now.

So, in summary, It seems that the total # connections on the hosts section does not account for everything. There were more connections in total on the overview section and it was failing from being overloaded.

Is this a problem or a misunderstanding on my part?

[ispyisail]

whats the router CPU/RAM load like?

[dsalch]

The CPU load was never very high at all. Almost idle.

The ram, I can't say that I noticed during times of trouble. It was not high enough to cause any red flags, but I can't say what it was.

[mgoo]

I think there is a problem, and not a misunderstanding on your part.

I have two wndrmacv2's, so when I decide to upgrade, I just make a swap of devices. Therefore I never keep any settings, and I never re-use any settings backups.

With the latest ispy april-15 exp release, I have experiensed similar behaviour, occasionally new connections are unable to establish, where already existing connections keep on going. I did not look at details like you did, I just guessed at some odd dns lookup issue with this particular release, or perhaps an error in my manually entered settings, and have since swapped the previously used wndrmacv2 back into operation. if anyone would want to have me look at specific settings or firmware info, I would be able to return to the ispy-15-apr, my limited time and effort permitting.

rgds
-mgoo

[Lantis]

If you have the maximum number of connections open and you try to open additional ones, they will fail. That is expected behaviour.

The connection number disparity I don't really understand where the data comes from but I'm sure it is explainable. They probably represent the same data in different ways.

[dsalch]

Here is an example of the situation I described earlier, I just cant make sense of these numbers...

Shouldn't the total # connections on the status screen equal the total of all the connections on the hosts screen?

Device
Device Name:Gargoyle
Gargoyle Version:1.9.2
Model:NETGEAR WNDR3700
Device Configuration:Gateway
Memory Usage:24.1MB / 59.6MB (40.4%)
Connections:5291/12288
CPU Load Averages:0.00 / 0.03 / 0.06 (1/5/15 minute

Hosts With Active Connections

• Hostname↓ Host IP Host MAC Active TCP Cxns Recent TCP Cxns UDP Cxns
  OfficePC-Wireless 192.168.1.25 70:F1:A1:FC:CA:38 703 853 73
  FamilyPC 192.168.1.26 90:E6:BA:32:81:D6 219 229 13
  IpObserver 192.168.1.27 00:0E:C6:00:01:DE 0 8 0
  BrendaPhone 192.168.1.73 88:79:7E:01:3B:7F 52 61 128
  ConnectMe 192.168.1.33 BC:38:D2:20:55:82 1 0 26
  DavePhone 192.168.1.72 88:79:7E:01:3B:88 2 0 2
  dsalch-ThinkPad-T420 192.168.1.130 74:E5:0B:9E:05:5A 2 0 0
  HP8620 192.168.1.4 9C:B6:54:5E:7C:8E 1 0 1
  dsalch-ThinkPad-X201-T... 192.168.1.196 58:94:6B:A1:63:DC 2 0 0
  dsalch-Lenovo-Yoga-2-Pro 192.168.1.124 7C:7A:91:75:F7:38 128 94 1

[mgoo]

seeing that youre using wndr3700, same hw series as my wndrmac's, I will try load the latest released exp ispy, and report here if my issue still persists

rgds
-mgoo

[mgoo]

after upgrading to ispy may-25 my particular issue seems to be gone.

-mgoo

[ispyisail]

after upgrading to ispy may-25 my particular issue seems to be gone.

-mgoo

The solution is more likely a fresh config.

One of the downsides to Gargoyle. Sometimes when you edit the config it can become damaged.

Fresh re-flash without using any backup file works wonders