An earnest appeal - please do fix APINGER in 2.2



  • Hi there,

    There has been numerous posts in various places in the forum on apinger issues that started appearing mostly from 2.1.x. These issues still exist in 2.2 Alpha (as of snapshot of couple weeks back).

    1. The main issue is that after some time (this period varies), the apinger shows wrong RTT and Loss numbers in the dashboard, or displays "Pending" and eventually excludes the gateway from gateway groups, while there's absolutely no actual RTT delay or packet loss in that WAN as tested by doing ping from any LAN client.

    2. Restarting apinger doesn't help much. In some cases within a minute of getting restarted, it reports the wrong RTT and loss and goes to "Pending" and then marks gateway as down (while in default gateway's case, pfsense is still using that gateway for pf to Internet traffic).

    3. The problem seems to be more prevalent for multi-WAN setups where the monitor IP is any Internet host with normal RTT of more than 50 ms. The problem lessens if the gateway IP is used as monitor IP (the default option) but then the value of testing whether the Internet connectivity is there or not diminishes.

    4. There's a weird workaround that isn't practical for everyone: if you have more than one WANs, use another router (like a wi-fi router) between the Internet connection and pfsense. Doing this, surprisingly stabilizes the apinger and it doesn't drift away much. In addition to this work-around, one can use a cron job entry to automatically restart apinger every hour or so, to further avoid apinger's wrong behavior.

    apinger is the basic mechanism required for smooth functioning of multi-WAN setups. So this is very critical that it works flawlessly. Many users are getting affected and not everyone is reporting. Many hours of productivity are being lost due to falsely making a gateway inactive. It should be fixed in 2.2.

    Here's a list of forum postings regarding various apinger-related issues:

    https://forum.pfsense.org/index.php?topic=68637
    https://forum.pfsense.org/index.php?topic=66328
    https://forum.pfsense.org/index.php?topic=69533
    https://forum.pfsense.org/index.php?topic=74914
    https://forum.pfsense.org/index.php?topic=72085
    https://forum.pfsense.org/index.php?topic=72314
    https://forum.pfsense.org/index.php?topic=76770
    https://forum.pfsense.org/index.php?topic=77266
    https://forum.pfsense.org/index.php?topic=73009
    https://forum.pfsense.org/index.php?topic=70441
    https://forum.pfsense.org/index.php?topic=72455
    https://forum.pfsense.org/index.php?topic=73109
    https://forum.pfsense.org/index.php?topic=72303
    https://forum.pfsense.org/index.php?topic=69261
    https://forum.pfsense.org/index.php?topic=71879
    https://forum.pfsense.org/index.php?topic=67354
    https://forum.pfsense.org/index.php?topic=68637
    https://forum.pfsense.org/index.php?topic=65505
    https://forum.pfsense.org/index.php?topic=63470.0

    Thanks,
    msu



  • (not being able to attach screenshot showing issue due to 500 internal server error). Here it is:

    http://imgur.com/GCqD1CV



  • I second pubmsu.
    I get more than 100 mails every day about these instable cable providers.
    At some days I get more than 2000 mails. All send within 5 minutes.



  • +1
    This is a quite annoying issue.



  • im also having this issue. i have 3 wans combined bandwidth but 2 of my gateways are always down. which is not. :(



  • They've had the gall to offer 4 general releases with this problem.  It makes anything higher than 2.0.3 worthless for anything more serious than a simple home router.

    Shame.  This is where software ideas like this go to die.


  • Rebel Alliance Developer Netgate

    The problem is that none of us can reproduce this on demand in an environment we control and can debug. Yes, we know some people have problems, but they don't affect the majority of users.

    Multi-WAN works fine here, and for many others. Ermal has been working on fixing this up but it's been a long process since the exact parameters to reproduce the problem have never been clearly identified or replicated.

    The notifications are a separate issue entirely, but that's not slated to be cleaned up on 2.2 but sometime afterward.



  • Thanks for your update jimp - we at least now know the difficulty in fixing it.

    If you want, I can help you with access to an otherwise good multi-WAN test environment that has this issue, which is an exact replication of our production environment. You can let me know in PM if this will help.

    We have been using pfSense for last 7 years I guess, and really need this to be resolved.


  • Rebel Alliance Developer Netgate

    Getting access wouldn't help as much as definitively identifying the specific condition leading to the problem if possible (e.g. a latency over X for Y amount of time, or Z gateways with Q latency, etc)

    Depending on how long it takes for the problem to repeat it could still be difficult for us to find time to watch it closely enough to find when the problem starts specifically.



  • Got it Jim. In our case though the problem starts within 5 to 10 minutes.


  • Rebel Alliance Developer Netgate

    Seeing the quality graph for your gateways may help as well, with notes about where apinger was restarted and where the problem was first noticed.

    I tried to artificially induce latency using one firewall in front of another and increasing the delay on a limiter for ICMP traffic. Each time I let it run for 15+ minutes at various latencies and then lifted the limiter. Each time it always bounced back to close to 0 for me, I never saw it get stuck, so there must be a few different factors at work making it get stuck over time for others.



  • Maybe this is something to consider: I never had any problems with my setup running NanoBSD for the last year. I switched to a full install recently (CF died, bought an SSD) and now I am seeing Packetloss steadily increasing for my HENet tunnel. 120% packetloss ATM, uptime of the firewall is 4 days. I am pinging the same IPv6 Host  via Smokeping from a Linux host behind the pfSense GW and the graphs look a little different. This is on 2.1.4, not on 2.2.








  • I have this problem too.  Apinger reports that my WAN connection keeps going up and down several times every hour.  It started a few months ago.  I have not switched ISPs or anything.  I installed the latest snapshot (built on Mon Jul 28 12:22:20 CDT 2014) and still have the problem.

    I do not use multiple WANs.  Just one.



  • I had this same issue as well and ended up coding in the local/private cable modems IP address into the config (192.168.100.1) and that was the workaround I used. Doesnt do anything for monitoring the connection but it's not always bouncing the connection up and down.



  • When you say "the config," do you mean the "Monitor IP"?  My config was monitoring the default gateway IP, which is on the cable modem and I still had the problem.



  • I don't think this is related to the main issue described here, but I have observed a similar behavior under high network load and while using the traffic shaper, because the ping probes are put on the default queue instead of the one specified by the floating rule on WAN that is supposed to handle the situation. Probably this happens because apinger starts before the firewall itself, since killing the related states makes them go into the correct queue immediately



  • Issue still exsits in recent bulids in my testing enviroments.



  • …and frequently results in tunnels (IPsec or openVPN) going down for no obvious reasons, except for apinger freakin' out.

    I increased the times for apinger alarm significantly, that helps at least a little...


  • Banned

    I dont have these problems at all running 40+ pfsenses….

    I use traceroute to monitor the wanted IP upstream to decide if the GW is down.

    All are stable currently running 0% packetloss..... No change from 2.0.X

    I dont like the idea of monitoring other external hosts not in your upstream environment. That way you dont get a real picture of your GW status.




  • @Supermule:

    I dont have these problems at all running 40+ pfsenses….

    What's your config for WAN interfaces? I see allot of people write that have problems but doesn't put configs to help troubleshoot the problem.

    Some times i have the problem in my multi-wan interface (PPPoE only config user and pass and a ppp (LTE) WAN only config default number). Don't see the problem when i disconnect my ppp.


  • Banned

    More or less the same for all 40+….




  • @Supermule:

    I use traceroute to monitor the wanted IP upstream to decide if the GW is down.

    All are stable currently running 0% packetloss….. No change from 2.0.X

    I dont like the idea of monitoring other external hosts not in your upstream environment. That way you dont get a real picture of your GW status.

    Could you please tell us how to use traceroute to monitor the wanted IP upstream to decide if the GW is down?

    Multi-wan with static IPs and different gateways within each wan subnets are stable at least in my tests,  but I use pppoe connections and we get the same gateway IP allmost all the times, so we have to set at least one monitor IP outside the wan subnet,  and this line with outside  monitor ip allways gets offline as the apinger reported, but the connection  functional as normal.

    If there is another to monitor the gatwway, that really helps.


  • Banned

    http://ping.eu/traceroute/

    Use the first one thats not in your WAN subnet.



  • @Supermule:

    http://ping.eu/traceroute/

    Use the first one thats not in your WAN subnet.

    It's not within pfsense, and not done  automaticly either?


  • Banned

    I understand why you are confused…

    I use traceroute to monitor the wanted IP upstream to decide if the GW is down.

    I use traceroute to locate the IP to monitor and then use the built in GW monitor tool in PFSense.

    Works fine here.



  • @Supermule:

    I understand why you are confused…

    I use traceroute to monitor the wanted IP upstream to decide if the GW is down.

    I use traceroute to locate the IP to monitor and then use the built in GW monitor tool in PFSense.

    Works fine here.

    OK, I did that several months ago,  and with no use.
    The next hop routers are always outside my wan subnet:(

    Thanks anyway.



  • I've been running into quite a few problems with apinger in the 2.1 series with a simple single wan configuration.

    https://redmine.pfsense.org/issues/3692

    I was thinking that I would try to debug it and went looking for the source code to apinger but wasn't able to find it in the 2.1.5 main or packages. Can someone send me a pointer to it?

    Thanks



  • @dennypage:

    I've been running into quite a few problems with apinger in the 2.1 series with a simple single wan configuration.

    https://redmine.pfsense.org/issues/3692

    I was thinking that I would try to debug it and went looking for the source code to apinger but wasn't able to find it in the 2.1.5 main or packages. Can someone send me a pointer to it?

    Thanks

    The source code is in the pfsense-tools repo.  You must complete a couple of electronic documents in order to access it.  Access is controlled via SSH public keys.  It is based off the FreeBSD port here: http://www.freshports.org/net/apinger/

    Information on what is required to get access is posted in a Sticky Thread at the top of the Development sub-forum here: https://forum.pfsense.org/index.php?topic=76132.0.

    Bill



  • @bmeeks:

    The source code is in the pfsense-tools repo.  You must complete a couple of electronic documents in order to access it.  Access is controlled via SSH public keys.  It is based off the FreeBSD port here: http://www.freshports.org/net/apinger/

    Information on what is required to get access is posted in a Sticky Thread at the top of the Development sub-forum here: https://forum.pfsense.org/index.php?topic=76132.0.

    Thanks. I downloaded what I thought was the packages distribution from

    https://github.com/pfsense/pfsense-packages/releases/tag/RELENG_2_1_5

    but apinger wasn't in there. Is there a different repo?



  • the packages repo is only for optional addons. (stuff that is not included in a base-install from CD)

    i believe bmeeks said in which repo you can find apinger, and how to access it



  • @heper:

    the packages repo is only for optional addons. (stuff that is not included in a base-install from CD)

    i believe bmeeks said in which repo you can find apinger, and how to access it

    Correct. It's in the pfsense-tools repo which is NOT hosted on Github directly.  It's hosted on a server operated by the pfSense team and you must follow the instructions in the link I posted above to gain access to the repo.

    Bill



  • I pushed a fix.

    Please try with next snapshots and see if you still get the same issues.



  • @ermal:

    I pushed a fix.

    Please try with next snapshots and see if you still get the same issues.

    Still broken for me. :(  See attachment.  Let me know if there is anything else I can provide.




  • Any update on the matter ?

    My box just went dark after apinger decided that the wan is not up and then remained stuck with a cpu load of 99% denying to be restarted.



  • APINGER does't cope well with my WAN flapping whenever this happens my HENET ipv6 gateway starts reporting loss forever even though the ipv6 tunnel is backup and stable!



  • why wouldn't adding some other methods of detection be a good idea had a xincom (same as syswan) that did 3 different means of detection (imcp ping, http heartbeat, traffic flow, DPD
    (RFC3706) ) years ago the was like a $250 router which was nice cause my isp use to block imcp



  • @naras:

    +1
    This is a quite annoying issue.

    and the outcome in my case is a nonfunctional apinger with miles of logentries like: apinger: No usable targets found, exiting



  • I'm having better luck with the 9/15 build.  I am not sure if something changed between 9/13 and 9/15 that would have affected apinger, but the status seems to be working better.  Still not 100% correct though.  Before, the connection would show insanely high RTT in the console; now, that isn't the case.  However, when I look at the logs I am getting a lot of apinger "down" log messages.






  • @applerule:

    I'm having better luck with the 9/15 build.

    Nevermind.  I've been checking it off and on and haven't seen it go "Latency" or "Offline" today.  It finally did.  Typical stuck at really high RTT and jumps between latency and offline.  I got too excited considering it hadn't gotten "stuck" for nearly a day.  :(




  • Still no luck with Sep 14 09:09:38 CDT 2014 build, two pppoe links are online before my isp resets my connections, after the reset, default gw gets online normally, but the opt1 gets stuck although the link reconnected successfully.

    And the OPT1 gets online again after I restart apinger.



Locked