1. If you're receiving a message that you are banned from the Current Events or Politics forums, it's not you specifically: those forums have been hidden for all users. For more info, see here.
    Dismiss Notice

ICS possibly causing data corruption

Discussion in 'Networking, Telephony & Internet' started by kinet1c, Aug 13, 2016.

  1. kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    OK guys long time no post, this is a 2 part piece, 2 problems, possibly connected... Scroll to the end for the 2 short version questions

    I havent lived in an area thats had technology for a while, but im in a nicer part of Fiji now and am in charge of a network set up by someone a long time ago.

    It is a hybrid nasty piece of work laid out over a resort to give most of the guests access to the internet and the 6 top shelf rooms have a smart tv with Plex running on them.

    The network is the following...

    - 3g unwired modem (yep... blistering speed for around 150 devices)

    - PC running windows 10 with Antamedia installed, ICS enabled (DHCP disabled) across the 2 NIC's both static IPs etc

    - TP-Link TL-WDR4300 Router running DD-WRT v24-sp2 DHCP enabled NAT disabled

    - PC running windows 10 and Plex media server connected to router

    - 2 TP-Link wa7210n and 1 wa5210g in access point mode from the router

    - 2 TP-Link wa7210n in Bridge with AP mode connected to 2 of the access points,
    1 to extend over the pool area to other rooms that have a router turned access point in each one daisy chained along from the Bridge and have a TV wirelessly connected in each room.
    The second to reach out to the staff areas.

    All fixed devices have static IPs outside of the DHCP range and using an I.P scanner and common sense can find no IP conflicts, the 3g modem and antamedia PC's WAN side NIC being on a different subnet. I can easily be missing a DNS or NAT issue somewhere, but I think its all ok, the internet DOES work, websites load, but files corrupt easily, previewing my own post one of the images displayed all greens and purples...
    All access point security is WPA2 AES and bgn 20mhz, except the bridges, they are n only and 40mhz.


    Click to view full size!



    The purpose of Antamedia is purely to throttle guests speeds to 25KB/s down and 16KB/s up and have a data cap of 250MB per day per MAC address (which no one reaches because of speed), chosen by the director. I would do away with it as DD-WRT im pretty sure has this option but the Director was some I.T wiz back in the day and loves this absolute pile of a program, so no budge there, it has the ability to charge people for use and redirect to social media for "likes" in exchange for use, he doesnt want this utilised (read idiot)

    The main problem I am trying to solve is small and random corruption of internet data of all kinds on any PC in the network except the one running Antamedia (direct connection to modem). The plex server is a very short link but experiences the same issues as a laptop connected out the back on the staff access point. Almost all downloads, .zip .exe .jpg etc will fail during download (network error) or complete, but then not unzip correctly, not run, or for jpg's, which will happen on a web page or after download will be scrambled or grey for most of the image. No PC's can complete windows updates or anything of the sort.
    One thing that works is P2P which Antamedia squashes anyway with timeout settings and the like, but there is a whitelist that staff MACs are added to so ive downloaded torrents with success which brought my line of thinking to protocol error.

    I've tried lots of different things and can provide info on each tactic but right now am looking for guidance from the beginning and fresh ideas so even the most basic suggestions are welcome so I can see that im not insane, yet.

    Im now looking at Internet connection sharing as the culprit for some sort of checksum type fault? (or something along the lines I have no real idea about this, thats where you guys come in!).

    2nd problem.

    So with this network design in mind some problems internally arise too as you can imagine. This is not internet related ( I don't think) but surprisingly when all devices feels like working You can stream 1080p mp4 files via plex to the furthest room in the network ~239 no problem, then other days the tvs wont connect to the server at all. This is very sporadic and the guests either dont complain about this too much or its rare so I dont get a real feel for failure rates here, on a pre-checkin room inspection plex will generally work. On the main PC running Antamedia there is Pinginfo view running to show if AP's need to be hard reset (we run from generators so power swap overs sometimes jam them up) look at these failure rates


    Click to view full size!


    Wireless channels for all devices are separated appropriately.

    I understand that pings aren't the be all and end all of network health and with so many devices behind that bridge it may not have time to answer the ping but the wired access point it connects to, 252, is always ready to say hello. Im probably missing something here too.

    As in the ping info pic there is a device at ~202, this is an access point in my room connected by cable to AP ~201, same setup as ~231 to ~234 and I have smb shares with the Plex server to watch movies and tv on a pc with Kodi, 90% of the time it works, the other 10% i would put down to the ~201 AP needing a hard reset.

    Would this ping failure hinder the network performance in any real world way? I can stream a file to the PC in my room and watch the ping failure climb across ~201 and ~202 but the video will play regardless.

    All network transfers are fine and will complete, not very fast, from my PC to the Plex server, when transferring around 200MB of files ill see 2-3MB/s, but no errors to speak of.

    So this is as much info as I can think of for now, ive been tinkering with this network for months, it has come a long way and it works, as in the guests can post selfies on facebook, i've even seen some using skype, apparently well enough.
    While I have access to the office network, with a separate, identical modem, get no corruption and slightly better speeds, the guests will complain the very second the network has a hiccup (we are 6 hours away from the mainland in the beautiful Yasawas, stare at the damn ocean!), so the higher powers arent too concerned about file corruption of images and zip files as long as the social media apps load so I am getting no more support on the matter. I just want to figure it out as peace of mind and a bit of learning. Im the manager/dive instructor of the Dive centre attached to the resort and find myself thinking of new things to try while underwater.

    It took me ages to type this and almost decided to wipe the lot with 2 simple questions posed as

    - Can Internet connection sharing cause data corruption?

    - Is a ping failure rate of 50% going to cause real world network issues for streaming videos?

    Now you have info to play with though.

    Thanks for reading guys and girls!
     
    Last edited: Aug 13, 2016
  2. evilasdeath

    evilasdeath Member

    Joined:
    Jul 24, 2004
    Messages:
    5,005
    interesting problem.

    Data corruption through a network to a host is VERY difficult to achieve. Devices will drop it the minute it fails FCS check, and even then the host will do CRCs on the packet as well and retry to receive it properly if it fails.

    Usually it's not so much errored packets, it's the end host being screwed and having a problem usually memory, it could be possible on the server doing ICS memory could be busted and it's rewriting the packets wrong (but correct to pass to the hosts) ie copy the packet wrong but write a correct FCS so it passes through the network.

    either way you are going to have to start ruling out devices, something is doing it.

    The other possibility is that something is just killing the TCP sessions early, ie antamedia (never used) or they are just timing out due to the tiny speeds

    just a theory.
     
  3. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    interesting indeed,

    I highly doubt its the end host computers ie my laptop, pc, the plex server, another laptop for a test, they all produce the same results. I hadnt thought of checking the server hardware though, might have to run a ram check late at night when its less likely there will be complaints of no internet.

    the speed isnt terrible, luckily the people in the resort have trips each day etc so right now i can do a speed test from my room and acheive this

    [​IMG]

    nothing to hang yourself over, i see download speeds at around 70-100KB/s but the end result is a corrupt file, if the download completes that is...

    example of corrupt zip


    Click to view full size!


    just now an example of image corruption


    Click to view full size!
     
  4. evilasdeath

    evilasdeath Member

    Joined:
    Jul 24, 2004
    Messages:
    5,005
    I'd guess both of them are caused by incomplete download complete not so much corrupted. What you have is probably correct it's just missing the end of the download
     
  5. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    Why do none of the browsers report an incomplete download then? they all have resume features, which i can use if the download fails half way or whatever, but sometimes the download will "complete" without interruption and in file explorer it will change from the browsers partial file extension to .zip etc, but then is a corrupt file.

    Same goes for images in webpages, you can refresh the page and it will stay the same, or get worse, or be fine the next time. Side note the browsers seem to rely less on cache these days with fast internet being the norm i guess?
     
  6. NiRdoC

    NiRdoC Member

    Joined:
    Jun 27, 2002
    Messages:
    200
    Location:
    Cairns
    I'd start with testing/replacing the cabling between the Antamedia workstation/server and the network.

    Alternatively for testing purposes bypass the Antamedia workstation/server and see if corruption still exists, you'll get an answer if it's the Antamedia workstation/server causing the issues then.

    Good Luck!
     
  7. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    See I for whatever reason never had isolation tests cross my mind, the guests will love it they will get "unlimited" speed for a little while...

    Cheers mate will report back later
     
  8. NiRdoC

    NiRdoC Member

    Joined:
    Jun 27, 2002
    Messages:
    200
    Location:
    Cairns
    Keep us posted on this one. It's a strange issue and I'm sure a few people who have read this thread would like the answer to. :thumbup:
     
  9. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    Over the last few days i tested 2 things

    1 - Disabling the Antamedia Server software and running the PC purely as an ICS machine, Data corruption remained, so I guess its the hardware of the Machine itself. How would I go about testing the hardware? RAM tests? Network card tests? Or could it even be drivers?

    2 - Using DD-WRT firmware I have the internet shaped for every I.P address allocated by the DHCP server and the corruption has dissappeared

    I thought the extra load on the router having to do all the work for every user on the network might cause issues but it has been stable since the re-configuration. My reasoning to the big bosses to allow the swap was with so many PC's all not receiving updates a virus may make its way into the network through 1 or more machines. I know this is very unlikely but it scared them enough to let the change stick.

    The network has no sensitive data on it from the company side so any loss would have only annoyed guests, which would not be good for us either way. Hopefully I can now concentrate on the massive ping issue internally plaguing the network.

    Any ideas there?
     
  10. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    Argh.

    The DD-WRT solution was not as elegant as I would have liked, the QoS settings seemed to work whenever they wanted, not very consistant at all. Sometimes each user would get a throttled speed of 25KB/s then suddenly the router would decide that everyone had to SHARE the 25KB/s across 150-200 devices, fairly slow...

    So I stripped the PC, cleaned all parts (Dusty, little to no thermal paste left) and with a fresh install of Windows 10 I have Antamedia running again. It controls speed much better, but the data corruption is there... Disabling the software so just ICS runs has the same result. I have found a few forums around complaining of the same issue with ICS but they only have dial-up connections, and the problem has appeared for them since moving from XP to Windows 7 (old posts). This may be an operating system issue.

    I am stuck on Windows 10 as the Antamedia software is licenced to that operating system install.


    What to do, what to do

    Edit. Have performed a ram test with memtest all good. Have set both network cards duplex settings manually at 10Mbps half duplex. No change. Should I look at ics settings like dns or anything like that? Totally out of my knowledge
     
    Last edited: Aug 23, 2016
  11. OP
    OP
    kinet1c

    kinet1c Member

    Joined:
    May 13, 2009
    Messages:
    68
    Location:
    Cronulla, NSW
    OK Update!

    I formatted the Server and installed windows 10 as UEFI, updated the BIOS, and didnt see a fix, although boot times are great...

    Found updated instructions for installing Antamedia on Windows 10, they now recommend turning off Windows Smart Screen, so thats done, I also found that ICS doesnt have uPnP on by default, and can only be turned on by enabling network discovery, which got my Playstation online which is neat! So, with all this in mind I now have no data corruption on any PC's. I am not sure if it was one or all of the changes but here we are.

    Another nice change was a new 4g modem and a plan change to Vodafone Fiji, much better speeds too!


    If someone can shed some light on which of the changes would have been the most benificial to the issue let me know, my advice to everyone else is keep away from Antamedia if you can help it.

    Cheers.
     
    Last edited: Sep 1, 2016

Share This Page

Advertisement: