The current version of the code is "fairly" stable, but not perfect.
I'm going to pick up a 5v N/C relay and powercycle the unit on a regular basis from another Propeller. This should prevent the "hang" states that it goes into from time to time. It'll be this weekend.
Please don't add the relay! That is, lets build some sort of logging and data collection into your code so we can find where the reason is that it hangs. I'd like to see the code fix and avoid the work around. At least built logging in so it writes data to an SD card or persistence means before each reboot.
This weekend I am trying to carve out two hours to work on exposing the RTR and RCR registers of the WIZnet, with this you should be able to set a longer timeout period. It may be the WIZnet is having trouble communicating with the people getting to the site.
The RTR (retry time register) is set at 200ms. If the W5100 doesn't get a response from the machine on the other end within 200ms it retrys. 200ms may seem like a lot, but when surfing from the opposite side of the world it doesn't sound like a lot. Any experienced web server professionals know of a similar value on a standard web server? I imagine we can turns the 200ms to 500ms or 1000ms (1 second) and see if the W5100 behaves better.
For the moment I am assuming there is something we can do with the W5100 driver for better performance, but it could just as very well be in non-driver code.
the retry time register night be the reason why I get an empty page most of the time (but not always) !
Where I live there is no wire-based fast internet-connection. So I bought a two-way satellite-based system. Works good so far only the ping-time is 700 - 1200 milliseconds.
Thanks Stefan for the data! I don't want to tout the RTR as the golden item to fix all, just as the first place I want to investigate, study and learn a bit more about. Your ping time is helpful in explaining one reason why I want to understand this parameter more.
I used PingTest.net from Singapore to Columbus, OH, I get the following results:
When I did a CMD line ping to OBC's device, I get results from 306ms to 968ms, sometimes no response, etc.
There may be other items to adjust too. I'll focus on the RTR this weekend and a little terminal for reviewing any status registers and parameters in the W5100. Then OBC could use my example, I hope, to pull data out and log the data. From there we can analyze and see what we learn.
When I
I made a ping from my PC to OBCs Spinneret Webserver
IP-Adress : 74.5.54.192 if this site Traceroute works right
Pingtime 850ms to 900 ms
I want to describe a little bit about this kind of internet-connection
this satellite-based systems seems to do a request which needs quite some time and then in most cases it makes pop! and the whole page is here.
I guess for small amounts of data the data is send at the highest possible speed the sat-modem can do.
When I use other computers webpages build up in 3 to 5 steps. What I get is a pause of maybe 3-5 seconds and then pop! the whole page is there
This is really great. I got it working from McMurdo, Antarctica using firefox and Safari. Very well done. Do you have any idea what top speed can be achieved on download between the uSD card and a client connecting through the server?
Comments
I guess if too many requests come in the spinneret gets crazy about that.
I don't know anything about programming a webserver
does there exist a possability to put requests into a que and just send a very short "busy - wait" message
instead of all the detailed data?
I'm going to pick up a 5v N/C relay and powercycle the unit on a regular basis from another Propeller. This should prevent the "hang" states that it goes into from time to time. It'll be this weekend.
OBC
This weekend I am trying to carve out two hours to work on exposing the RTR and RCR registers of the WIZnet, with this you should be able to set a longer timeout period. It may be the WIZnet is having trouble communicating with the people getting to the site.
The RTR (retry time register) is set at 200ms. If the W5100 doesn't get a response from the machine on the other end within 200ms it retrys. 200ms may seem like a lot, but when surfing from the opposite side of the world it doesn't sound like a lot. Any experienced web server professionals know of a similar value on a standard web server? I imagine we can turns the 200ms to 500ms or 1000ms (1 second) and see if the W5100 behaves better.
For the moment I am assuming there is something we can do with the W5100 driver for better performance, but it could just as very well be in non-driver code.
the retry time register night be the reason why I get an empty page most of the time (but not always) !
Where I live there is no wire-based fast internet-connection. So I bought a two-way satellite-based system. Works good so far only the ping-time is 700 - 1200 milliseconds.
best regards
Stefan
I used PingTest.net from Singapore to Columbus, OH, I get the following results:
When I did a CMD line ping to OBC's device, I get results from 306ms to 968ms, sometimes no response, etc.
There may be other items to adjust too. I'll focus on the RTR this weekend and a little terminal for reviewing any status registers and parameters in the W5100. Then OBC could use my example, I hope, to pull data out and log the data. From there we can analyze and see what we learn.
When I
I made a ping from my PC to OBCs Spinneret Webserver
IP-Adress : 74.5.54.192 if this site Traceroute works right
Pingtime 850ms to 900 ms
I want to describe a little bit about this kind of internet-connection
this satellite-based systems seems to do a request which needs quite some time and then in most cases it makes pop! and the whole page is here.
I guess for small amounts of data the data is send at the highest possible speed the sat-modem can do.
When I use other computers webpages build up in 3 to 5 steps. What I get is a pause of maybe 3-5 seconds and then pop! the whole page is there
If you want meto do some testing let me know
best regards
Stefan
This is really great. I got it working from McMurdo, Antarctica using firefox and Safari. Very well done. Do you have any idea what top speed can be achieved on download between the uSD card and a client connecting through the server?
Again, very well done.
Peter