CME error 558 and sessions

7 thoughts on “CME error 558 and sessions

  1. Hi,

     

    One of our providers DNS is not working properly and this seems to be causing session issues.

     

    The following sequence takes place:

     

    at#sd=1,0,80,"www.example.com",0,0

     

    +CME ERROR: 558

     

    at#sd=1,0,80,"www.example.com",0,0

    CONNECT

     

     After the initial DNS error the socket dial is immediately repeated (one or more times) and contact is made. From billing analysis it appears that the above DNS errors somehow result in connections not being closed properly. Should a #sh be issued before retrying the socket dial or is there some other issue to consider here?

     

    update: Issuing at#sh doesn’t help, this seems to be a session issue resulting from the DNS error.

     

    Any help would be greatly appreciated,

     

    Thanks,

    Wiebe

    1. To go around that particular DNS problem, test its reliability with AT#QDNS; cache the result either automatically using AT#CACHEDNS or do it yourself and open the socket with the IP instead of address; use another set of DNS servers, see AT#DNS.

       

      1. Hi Cosmin,

         

        Our provider appears to be seeing the same DNS session issues so it doesn’t look like the problem is related to the AT sequences I’m using.

         

        Thanks for the useful caching suggestion. I noticed that the TTL on many DNS’s is often quite short, can a longer period be selected for  #CACHEDNS?

         

         

        Kind regards,

        Wiebe

         

         

         

         

         

         

      2. Hi Cosmin,

         

        Our provider appears to be seeing the same
        DNS session issues so it doesn’t look like the problem is related to the
        AT sequences I’m using.

         

        Thanks for the useful caching
        suggestion. I noticed that the TTL on many DNS’s is often quite short,
        can a longer period be selected for  #CACHEDNS?

         

         

        Kind regards,

        Wiebe

        1. I think TTL is set  by the administrators with good reasons, their update rate and other propagation issues, operational procedures etc. I would say that is not wise to have a longer cache time, while having a shorter one is good for dynamic DNS users.

           

          1. I didn’t mean changing the TTL of the DNS. My understanding of #CACHEDNS is that it copies the TTL of the DNS for local use. I’d want to cache the IP for several days due to infrequent transmissions but this wouldn’t work with #CACHEDNS given the typically short TTL’s used.

             

            My question was whether you could set your own TTL for #CACHEDNS to avoid having to write your own cache (which is what #CACHEDNS effectively is).

             

            Thanks for the feedback,

            Wiebe

          2. Probably the only way then is to keep the name/IP mapping yourself in a variable somewhere.