Topic: Dreams becoming reality for the D2: GW4 MySQL test

[RazalYllib]

It is my opinion that sql should work.  The last true stress test was several dozen people, and IIRC they were all in the same hex and someone selected a mission.  Until that point in time, the server was doing OK, bit of a stutter, but ok.  The stutter was do to the logging.  I strongly urge the powers that be, to use the sql server kit when GW4 is live.  There may be absolutely wrong with it.  The problem experienced in the past may not be present during the course of normal play.  I would use the sql kit, and not tell the player base...ala single blind experiment.  If it crashes after a couple hours then restart with flatfile, but I do not think it will crash.

[RazalYllib]

Just curious, was enough data gathered to make worthwhile????

[FPF-DieHard]

I guess my problem with using a major server as a stress test is what happened with RDSL (was that the one??).  After crashing like nuts, it had to go to a flatfile after a number of days, I believe.  I'm just hoping that you all have the flatfile ready to go if the SQL craps out.  If it works, great.  If not, there are going to be alot of frustrated people.   Let's hope for a bit of luck.

There was a single 20 minute SQL test for RDSL thats it, then RDSL was on the flatfile and its issues were gf settings (hourly burps). The kit's problems go waaaaayyyyyy beyond just how it handles SQL. The kit does not report gf errors in a useful way, the settings that are in error can only be determined by trial and error - lots of fun - give it a try sometime!

I'll bet you a week's pay the issue on SQL was donor ships in the Server-side shiplist.   I removed all the donor ships from the server list for GW and things seem to be relatively stable.   GW3 was more stable, but there are other variables to account for that.

[Bonk]

Just curious, was enough data gathered to make worthwhile????

Yes we gathered enough data to make some conclusions about the serverkit (whether under load or not and whether on SQL or not). It turned out not to be a stress test at all but nevertheless was very revealing. Thanks to all those that helped!

[RazalYllib]

Reveal what was revealed

[Bonk]

Reveal what was revealed

The majority of SQL tests that have failed have had nothing to do with SQL and everything to do with undetermined gf errors and shiplist misconfigurations.

[RazalYllib]

Well that must be progress eh?