PRPNet 3.0.0 Released

[henryzz]

Originally Posted by MyDogBuster

Geez, doe's that leave me as the only retired one of the bunch. Now I really feel old. Heck, I retired (1995) before some of your kids were born.

I was only 3 or 4.
Makes me feel young lol

[rogue]

Originally Posted by MyDogBuster

Max is correct. Although on my test, half the screen was unreadable.

It did look like the user screen though and not the stats screen I was going for.

BTW, the new table userprimes was populated correctly on my test. 176 entries and they look just fine.

Can you send me the HTML that the server sent? It might be somewhere in your cache if you can't reproduce it.

The server should display "Page does not exist on server" if you either don't specify a page or if the page specified is not configured. Are you certain that the browser did auto-finish the page name when you entered the "/"?

[rogue]

Originally Posted by MyDogBuster

Finally got a hit. Debug files emailed. I forget to put in the email that it's from 3.1.5 not 3.2.0.

I got your logs. Normally GETWORK returns the first candidate in about 5 seconds. In the case where it failed, it took 18 seconds. The client gave up waiting after 10.

I did make some changes in 3.2.0 that should help with this problem. The first was to have the client wait for up to 60 seconds. The second was to have the server limit the result set to 1000 records. When large amounts of data or retrieved from the database, MySQL puts them into memory. Putting them into memory takes time. By reducing the number of records in memory, the cursor will open much faster.

There are two possible points of failure with these changes. First, that the server is so bogged down that it takes more than 60 second to open the cursor (even with the limit). Second, that the server is configured to do double-checks, but the clients that are connecting cannot double-check any of the first thousand records.

I have a fix in mind for the first problem, but I need to think about it more before implementing it. I also have a fix in mind the the second problem, but that is much lower in priority and would require the first fix to be done.

Would you feel comfortable trying 3.2.0 and seeing how much the time lag decreases in the server between getting a GETWORK request and responding to it?

[MyDogBuster]

Would you feel comfortable trying 3.2.0 and seeing how much the time lag decreases in the server between getting a GETWORK request and responding to it?

Couple of things in response. I did try 3.2.0 and got the message with 1 client and the server running on the first attempt to return results. I got the stats html page failure shortly after that and went back to 3.1.5.

Right now I'm digging out from 5 feet of snow over the last 6 days. When I get a chance I'll give 3.2.0 another shot. I did however have the client timeout set to 1 minute (I think). Question, the errortimeout setting in the client config file, is it in minutes or seconds or what?

Secondly, I sent you a Firefox page of the webpage bug. If you can't read that, I have it saved as a jpeg image (screenshot). Let me know if you need that.

[rogue]

Originally Posted by MyDogBuster

Couple of things in response. I did try 3.2.0 and got the message with 1 client and the server running on the first attempt to return results. I got the stats html page failure shortly after that and went back to 3.1.5.

Right now I'm digging out from 5 feet of snow over the last 6 days. When I get a chance I'll give 3.2.0 another shot. I did however have the client timeout set to 1 minute (I think). Question, the errortimeout setting in the client config file, is it in minutes or seconds or what?

Secondly, I sent you a Firefox page of the webpage bug. If you can't read that, I have it saved as a jpeg image (screenshot). Let me know if you need that.

I got your e-mail. I see the stupid mistake I made. I'll fix it in 3.2.1.

As for the first issue you mentioned with 3.2.0, the debug logs would be really helpful. I don't think it is the same issue that you experienced in 3.1.5.

Enjoy your snow. We just got a foot of the stuff Tuesday. My boys are looking forward to going tubing.

[MyDogBuster]

Question, the errortimeout setting in the client config file, is it in minutes or seconds or what?

Mark, you missed my question.

[rogue]

Originally Posted by MyDogBuster

Mark, you missed my question.

Sorry about that. It is in minutes.

[Xentar]

Originally Posted by rogue

Data should not be lost. Did the violation happen in the server or the client? If it happened in the server, then it might be the same crash that I experienced.

It happened in the server.
I enabled debug logging, but the problem didn't occur again..

[rogue]

Originally Posted by Xentar

It happened in the server.
I enabled debug logging, but the problem didn't occur again..

Did the problem occur right away or did the server run for a while before it crashed? Do you have Visual Studio installed on the box? If so, you could build the server yourself. When it crashes it would start Visual Studio and tell you where it crashed.

[gd_barnes]

Originally Posted by henryzz

I was only 3 or 4.
Makes me feel young lol

My son is near your age. He just turned 17. lol My daughter is 14. Born in '93 and '95.

Don't let Ian fool you guys too much. He's not totally antient, just rich. (lol) He may have retired 15 years ago but I think he was in his early 50's. :-) I'm pretty sure he's younger than my parents now (71 & 70).

Hope I didn't give away the farm there Ian. :-)

[MyDogBuster]

Hope I didn't give away the farm there Ian. :-)

Right on. I retired in 1995 and I'm 64 years old.