w-u-2-o wrote:Richie,
After about 8 hours of continuous receive operation and three short QSOs, without paying any attention to checking the log, this is what I have at the end of the day.
It might be handy if the log had time tags and was also tagged as to whether the mode was RX or TX.
To me it looks like a packet reorder buffer with a depth of 5 would have cured most of these ills, except for those instances of 1512--am I understanding this correctly?
Thanks!
Scott
Seq Error Log.txt
Hi Scott,
Any +ve number will be the packet arriving has a sequence number greater than expected. As you are not seeing any -ve numbers at all, then there are no packets arriving out of order, with a lower sequence number than expected. This implies loss only.
For example, if you take the 4's. The packet arriving is 4 ahead of expected, however, the next (or subsequent nexts) do not see any packets that were prior to that higher than expected. Remember, in the code the next packet it is expecting will now be the one after the erroneous 4.
As soon as an out of sequence number is detected a snapshot is made, so you will not get the next packet number included. However, if the next packet is out (ie not the one after your changed sequence number) then another snapshot will happen instantly and you will see two indicated packet sequence errors.
It is hard to explain, but unless you see -ve numbers you are not really getting out of order packets (ie old packets arriving after they should have). As far as I read your log, you are getting small packet loss, except for the 1512's and 245's which are somewhat larger.
BTW, if I use 'clumsy' and turn reorder on, then loads of -ve numbers appear.
edit: yes about time tags and tagging mox state
edit2: and for what it is worth I see similar 200>1000 sequence number errors here, also implying times of heavy loss
Richie.