Test Server vs Live Servers [Archive] - EQ2 Forum Archive @ EQ2Wire

Dyeana

04-10-2005, 04:58 AM

<DIV> Test Server vs Live servers After a while, all the complaints about the Test server letting through another bug or someone today starts up another “Let’s Copy” campaign tend to get old fast, becoming in themselves just more spam. I wish the board manager could place a filter on them for a fast self delete. How about at least once coming over to Test Server for a few weeks, and really learn how they have to do the large patches and other changes? And find out what also is involved in testing and working out bugs that can be identified on Test server. Then compare live servers to test servers. Don’t have the time for another character, then I doubt “Copy” would work either, You would drop in long enough to get the exploit or if lucky make an observation and bug report or two, then play your regular server character to support your guild or your enjoyment. And yes, copy is used in some testing of content but not random uncontrolled copy. My main characters I play are on Test server, but I do go to the live servers to visit the dark side and see how changes which I saw parts of actually do get used and some pointers I can take back to my main test server characters to improve them. With the sheer numbers of people on live servers, so much more is possible faster because I may be able to bypass having to kill some mobs or may more readily get help or pointers for the large number of other players. I sort-of look at the live servers as my test area and Test server as my more challenging live area. Here is an example, (and I am probably very much over simplifying their actual process flow just for the sake of an example) placed in layman’s terms as much as possible, of what is transpiring; Let’s say 10 to 14 days before a major patch, the developers are essentially complete, the code is debugged on the debugging test server, and ready for the changes to be applied to the Test Server. For this example, we’ll use 14 days before the big “Alpha” patch since we know the Dev’s are on the ball. So you pull in a clean verified copy of the “Live Servers”, patch the change What is the worse that can go wrong? Well when our top notch Dev was developing the changes, he was working with a clean copy of the Test server or live servers a few weeks old already. As with any product in a constant flux to optimize performance, changes were made in the interim of the Dev getting the new mater ready. So another debug is in order to find the changes needed to make the patch to the Test server. So now the product is patched 14 days prior to when needed. The next thing which is necessary for the testing of the patch is a statement to others our top notch Dev is working with to “Hands off” the Test server, meaning make no changes so testing can take place is a somewhat controlled atmosphere and changes needed can be made to the Test server. Sure, a couple changes may be made, but for the most part, if it is not a required Test server change or doesn’t affect the coded area of the patch, no changes are made to the Test server. During this testing period, what else is transpiring, ….. well our super genus “Yantis” types decided they found a new easy exploit for which a macro can be developed, Cousin Sue got 15 plat for her rare level 3 bracelet which no other players decided was worth their effort for the last 2 months, Billy-Bob decided they wanted to see the screen in a different manner and changed options causing his screen to blank out and options to not reset since he decided to reset his computer in the meantime, or Grandpa completed quest XYZ for the first time on “ABC” server and causes a zone or server reset due to a glitch in programming. Some bug reports or feedback reports may have been made and a Dev takes a look at it. If serious, the live servers are brought down immediately, if medium, a change is made for the next daily update, and if not game threatening, it may wait for the next large patch or beyond. Well, changes are made to the Live Servers, patches installed during the morning server reset, and other tweaks made to ensure optimum performance to the game servers anticipating a certain amount of loading through-out the day. With the massive amount of people logging on and off the servers through out the day, the number of memory leaks occur in uncompleted actions and other ques still waiting on the servers. EQII being very memory intensive, optimum performance has almost dictated a daily reset to clear these out and start fresh for another day. Sure they can go much longer, but there are less problems in live servers getting out of sync and scheduling synergies with doing them daily. The Test server never gets these changes for the most part. There can be one to two dozen changes or more made to live servers which are not applied to the Test server being tested. A day or so before the big alpha change is to be made, the patch is prepped for the live servers the next day. The grand day arrives, patch installed, live servers reset, and the individual live servers verified and brought back online. Another EQII day begins. …. Followed shortly by an all servers down, sometime due to the many changes which took place during the 2 weeks which were not accommodated for in added memory locations or cousin Vinny looked at the patch notes and daydreamed while waiting to get online, that he could try this and that now. All sorts of weird gyrations get tried the day the patch goes live .. come on … we have to find out what our new limitations really are … right? So the server went down, you are waiting for it to come back and are bored at you 300 mile an hour mind racing to be the first to do this or that on the server. In the sheer boredom which seems like days although is was an hour or less, you see this nifty area to post comments on testing or cloning and post another comment or two. Well, unfortunately, Test server never saw all the changes made to the live servers due to it’s nature as a test server. And with the large number of changes in one of these large patches, even having the heavy loading of a live server will not identify a problem arising before it is patched to the live servers. New code usually goes to the bottom and uses routines and gaps in preceding code to the max extent possible. Some live changes add code t the area intended for the patch or alter the goto’s in some very unexpected manners. Take a couple language and programming computer courses and you’ll understand quickly how a program evolves, and might get involved in program management. About the only thing I can identify for SoE relating to patching is to try and have all changes made to the live servers and the big patch, made and placed on the test server the day or two before the scheduled date, and do a freeze on live servers longer or as long as possible before the change. Might be nice if there were not the constant daily changes being made so you can have a more stable platform, but circumstances are not always optimum and if is difficult keep up with the “super geniuses Yantis types” who wish to create macro’s and other related programs for fun or profit. The only area for improvement is Change Management as difficult as it seems. But the guys do a pretty good job at Change management seeing the amount of change that really takes place. I tip my bonnet to you guys! Dye “The few, the proud, the Test Servers! Are you up to the challenge?”</DIV>