• Problem found with downlinks heavily polling

    From Vince Coen@2:250/1 to All on Sun Sep 8 16:58:34 2019
    Hello All!

    I have noticed that for downlinks polling more often than once per hour where the echo packets extention have gone (for example) from :

    0000fffc.su0 to .su9
    0000fffc.sua to .suz

    The system can no longer pass packets to them as all extention letters are used
    up.

    Here we have the above archives all zero but the system does not clean itself up of empty archive packets after they have been sent so in the above example with 36 archives all empty mbse has no where to go for this node and s/he just looses any more that might arrive for them.

    This is caused by every time a node polls any archives on hold are sent (so
    far so good) with mbse zeroising the size of each that has been sent.

    On the next poll it creates a new extention preset to zero and if by the next poll it is still zero will still create a new extention and not just hold as is.

    The other factor would be having reached say .suz to start with suA thru suZ but here I am guessing that some systems may well not be able to cope with the upper case name ?

    Another option is if mbse on reaching .xxz (where xx is the day) to delete all older empty archives and then restart with .xx0.


    Any one with ideas on this issue ?


    And yes I have checked that in this instance for the above archives (0000fffc) they all are created at time of polling by the node.


    Vince

    --- Mageia Linux v6 X64/Mbse v1.0.7.12/GoldED+/LNX 1.1.501-b20150715
    * Origin: Air Applewood, The Linux Gateway to the UK & Eire (2:250/1)
  • From mark lewis@1:3634/12.73 to Vince Coen on Mon Sep 9 09:00:28 2019

    On 2019 Sep 08 16:58:34, you wrote to All:

    The system can no longer pass packets to them as all extention letters
    are used up.

    this is one reason why these days it is easier and somewhat better to not use bundles at all... just raw PKTs and there's no more problem like this specific one...

    Another option is if mbse on reaching .xxz (where xx is the day) to delete all older empty archives and then restart with .xx0.

    this is what i remember being done BUT when the new bundle was created, the old
    truncated one was removed so it appears that your problem is in the code that should be removing the last truncated bundle one the new bundle's name has been
    determined...

    of course, if you insist on using bundles, you could drop the extension increments and always send

    .mo1
    .tu1
    .we1
    etc...

    that might be done by setting your system to delete the bundle instead of truncating it...

    i switched to raw PKTs over a decade ago and have never looked back...

    )\/(ark

    Once men turned their thinking over to machines in the hope that this would set
    them free. But that only permitted other men with machines to enslave them.
    ... The tin can was invented in 1810; the can opener in 1855.
    ---
    * Origin: (1:3634/12.73)
  • From Vince Coen@2:250/1 to mark lewis on Mon Sep 9 15:32:45 2019
    Hello mark!

    Monday September 09 2019 09:00, you wrote to me:


    On 2019 Sep 08 16:58:34, you wrote to All:

    The system can no longer pass packets to them as all extention
    letters are used up.

    this is one reason why these days it is easier and somewhat better to
    not use bundles at all... just raw PKTs and there's no more problem
    like this specific one...

    Another option is if mbse on reaching .xxz (where xx is the day)
    to delete all older empty archives and then restart with .xx0.

    this is what i remember being done BUT when the new bundle was
    created, the old truncated one was removed so it appears that your
    problem is in the code that should be removing the last truncated
    bundle one the new bundle's name has been determined...

    of course, if you insist on using bundles, you could drop the
    extension increments and always send

    .mo1
    .tu1
    .we1
    etc...

    that might be done by setting your system to delete the bundle instead
    of truncating it...

    i switched to raw PKTs over a decade ago and have never looked back...


    MAy well have to do the same but looking yesterday at the last archive used late afternoon it does appear that mbse may well be using the last one for new packets. I will keep an eye on it over the next few days but today it looks like the downlink has reduced his polling count over a 24 hour period to once per hour but must remember to check it after 23:00 to verify.


    Vince

    --- Mageia Linux v6 X64/Mbse v1.0.7.12/GoldED+/LNX 1.1.501-b20150715
    * Origin: Air Applewood, The Linux Gateway to the UK & Eire (2:250/1)
  • From mark lewis@1:3634/12.73 to Vince Coen on Sat Sep 14 14:36:42 2019

    On 2019 Sep 09 15:32:44, you wrote to me:

    i switched to raw PKTs over a decade ago and have never looked back...

    MAy well have to do the same

    from my side of the fence, it is recommended... especially if you're in a "crash-me-crash-you" delivery setup where all mail is delivered ASAP once it is
    processed and ready to go...

    )\/(ark

    Once men turned their thinking over to machines in the hope that this would set
    them free. But that only permitted other men with machines to enslave them.
    ... Torontonians can't spit out the car window without stopping.
    ---
    * Origin: (1:3634/12.73)
  • From Vince Coen@2:250/1 to mark lewis on Sun Sep 15 13:30:09 2019
    Hello mark!

    Saturday September 14 2019 14:36, you wrote to me:


    On 2019 Sep 09 15:32:44, you wrote to me:

    i switched to raw PKTs over a decade ago and have never looked
    back...

    MAy well have to do the same

    from my side of the fence, it is recommended... especially if you're
    in a "crash-me-crash-you" delivery setup where all mail is delivered
    ASAP once it is processed and ready to go...

    Nope, all mail/files are put on hold for downlinks.

    If it gets to be over 7 days since a poll (outside July, August and Dec, Jan) attempt to poll a few days running if that fails system puts the node on pause passes me a msg to put them as hold in the nodelist segment. After 14 days they
    go as down and 21 days they are remarked out.

    Sometimes (must get around to looking at that) my system will notify me that they have polled me after this and I will remove the 'remark' char in the nodelist and unpause them for mail/files.

    Must figure out how to fully automate this all some day.

    Vince

    --- Mageia Linux v6 X64/Mbse v1.0.7.12A/GoldED+/LNX 1.1.501-b20150715
    * Origin: Air Applewood, The Linux Gateway to the UK & Eire (2:250/1)