Bug #3191: Quality RRD inaccuracies and failure to update status in some circumstances - pfSense - pfSense bugtracker

Actions

Copy link

Bug #3191

closed

Quality RRD inaccuracies and failure to update status in some circumstances

Added by Chris Buechler over 11 years ago. Updated about 9 years ago.

Status:

Resolved

Priority:

Normal

Assignee:

Chris Buechler

Category:

Gateway Monitoring

Target version:

2.3

Start date:

09/09/2013

Due date:

% Done:

Estimated time:

Plus Target Version:

Release Notes:

Affected Version:

All

Affected Architecture:

Description

There are some circumstances in which apinger puts in incorrect quality RRD data and fails to notice downtime. This needs some more specifics, but here are known examples of circumstances that fail.

1) use a monitor IP with 80% packet loss, with "Down" at the default 10, and it won't detect any packet loss and graphs no packet loss.
2) use a monitor IP with 80% packet loss, with "Down" at 30, and it'll take the gateway down appropriately but the RRD data will be wrong. Shows 100% loss much of the time.

There may be other specific circumstances that need to be added here.

This is true across all versions that have ever used apinger.

Actions

Copy link

Updated by Ermal Luçi almost 11 years ago

Target version changed from 2.1.1 to 2.2

I am not sure this is something to be fixed for 2.1.1 so putting to 2.2.

Actions

Copy link

Updated by Jim Thompson over 10 years ago

Assignee set to Chris Buechler

need more info on 'when' this happens (why would be great)

Actions

Copy link

Updated by Chris Buechler over 10 years ago

Assignee changed from Chris Buechler to Ermal Luçi

There are a few descriptions of problems in tickets in Kayako under the apinger-badstats tag.

Actions

Copy link

Updated by Ermal Luçi over 10 years ago

Status changed from New to Feedback

Patched apinger, need some feedback if the issue is solved now.

Actions

Copy link

Updated by Ermal Luçi over 10 years ago

For the record, properly recover from disconnected sockets patch put in.

Actions

Copy link

Updated by Chris Buechler about 10 years ago

Status changed from Feedback to New

The first issue as noted originally is still a problem as described. Throw a limiter on an upstream system that drops 80% of the monitor pings, and the packet loss and latency reported by apinger are completely wrong. Reports somewhere around 3 ms with 0% loss when in reality, reading back a pcap with tshark:

ICMP Service Response Time (SRT) Statistics (all times in ms):
Filter: ip.addr==192.0.2.202

Requests  Replies   Lost      % Loss
38        9         29         76.3%

Minimum   Maximum   Mean      Median    SDeviation     Min Frame Max Frame
90.927    94.974    92.275    92.089    1.379          54        34

Its pings are getting replies in ~92-94 ms (not the 3 it's showing), and the loss doesn't show up at all.

Change "Down" to 30, and the RRD data is correct for loss, but not latency.

Actions

Copy link

Updated by Chris Buechler about 10 years ago

Assignee changed from Ermal Luçi to Chris Buechler

to me to re-test

Actions

Copy link

Updated by Chris Buechler about 10 years ago

Status changed from New to Feedback
Target version changed from 2.2 to 2.3

things are much better with apinger in general after fixes in the past 1-2 months. I can still replicate some issues here, but it's unlikely they're ones that would occur in the real world. Given things are better than in current stable releases as is, and touching apinger is fraught with peril, I'm setting this out to 2.3 for feedback and further review.

Actions

Copy link

#10