ESET NOD32 RA HTTP Server stops responding...

Discussion in 'Other ESET Home Products' started by questorz, Aug 3, 2009.

Thread Status:
Not open for further replies.
  1. questorz

    questorz Registered Member

    Joined:
    Feb 3, 2006
    Posts:
    6
    We are using NOD32 v4. Our clients connect back to an RA server (3.1.11)that the HTTP update server service (ESET RA HTTP Server) stops responding on port 2221. Restarting the service seems to fix the problem temporarily. This happens multiple times a day and is driving us crazy. Has anyone else experienced this? The service actually doesn't shut down (to where it would be automatically re-started) and there are no errors in the windows event logs...

    Thanks,
    --Mike
     
  2. Marcos

    Marcos Eset Staff Account

    Joined:
    Nov 22, 2002
    Posts:
    14,374
    How many clients connect to the server?
     
  3. questorz

    questorz Registered Member

    Joined:
    Feb 3, 2006
    Posts:
    6
    Only 50-60 at the moment. We had over 3000 on a single server with the version 2 series. I am using mysql for the log database. That's the main difference in the setup. During this time the RA function where the clients check in, pull configs, etc, is functioning perfectly. So that service seems OK.
     
  4. SmackyTheFrog

    SmackyTheFrog Registered Member

    Joined:
    Nov 5, 2007
    Posts:
    767
    Location:
    Lansing, Michigan
    I ended up moving over to IIS to host my mirror to resolve an issue with client connections not closing properly and spawning a bunch of zombie threads under the RA HTTP service. It ended up working really well for me and would recommend moving over to that configuration if it is an option for you.
     
  5. questorz

    questorz Registered Member

    Joined:
    Feb 3, 2006
    Posts:
    6
    Are you running IS on the same server that the ESET server is running on and just pointing a public directory to the area the mirror is stored at? Do you do authentication also?
     
  6. Brambb

    Brambb Registered Member

    Joined:
    Sep 25, 2006
    Posts:
    411
    Location:
    The Netherlands
  7. SmackyTheFrog

    SmackyTheFrog Registered Member

    Joined:
    Nov 5, 2007
    Posts:
    767
    Location:
    Lansing, Michigan
    To answer both of your questions, "Yes." IIS is serving the local mirror directory as its HTTP root and I enabled IIS authentication so clients have to give the credentials for a user account on the system which is authenticated over NTLM. If RAS is on a domain then you will either need to use a domain account or jump though some hoops to get IIS to authenticate against a local account.
     
  8. mastj25

    mastj25 Registered Member

    Joined:
    Apr 20, 2009
    Posts:
    22
    Did you ever figure out a fix to this? I have just started test with Version 4, clients are version 4.0.467 and have been experiencing this same issue the past few days.
     
  9. matrak

    matrak Registered Member

    Joined:
    Sep 16, 2009
    Posts:
    11
    I confirm that using the IIS on ERAS server works fine. Now , I have 500 clients :)

    I think the RA HTTP server only a certain number of customers is mismanaging the socket openings...A bug ?
     
  10. bio

    bio Registered Member

    Joined:
    May 22, 2007
    Posts:
    9

    same problem here.. will migrate to IIS as well this week.... nod v2.7 worked fine with 900 connections o_O

    bio..
     
  11. jreiter

    jreiter Registered Member

    Joined:
    Sep 22, 2009
    Posts:
    28
    I have to agree with everything SmackyTheFrog said. I ended up migrating our update server to IIS, and it has worked perfectly ever since. The ERAS HTTP service is simply broken. We started seeing the zombied connections after only 50 clients were hitting it, and it only got worse as we rolled out all 1000 of our clients.

    So for now, just use IIS to serve your updates. It's faster, more reliable, and honestly, more configurable and more powerful.
     
Thread Status:
Not open for further replies.