Problems With Thai Servers?

[newsite12]

Here is a total mind bender for you all.

We have been selling Thai penny auction scripts for the past 6 months. All using the same core php and mysql scipt.

About 6 to 8 weeks ago, the bidbutlers started going a bit funny, as in, the would run quite normally up to 1000 to 2000 baht, and then just randomly decide to kill themselves and stop bidding and all the listings would end. We can't understand why it is happening. We normally install on Hostinglotus or hostneverdie. and run on VPS with CentOS.

We have had 5 coders look at it, and none can find a problem in the coding and all have said that it seems to be a server time issue. which the hosts flatly deny.

Strange thing is, that we looked back at all the past clients site and the problem has occurred on all but one website.... and the website that is running OK, is one that was put up on dedicated servers. Which really confirms server issue.

Although technically it is not our fault, and that we are not really responsible for client's server issues, and that we delivered a fully working website when money changed hands. But we really would like to clear up this mess, as we want everyone to be happy.

Has anyone heard of (or know of) any changes that were introduced to the major hosts in thailand?

There has to be something to do with it.

Thank in advance for any help.

[Crushdepth]

Exactly what are these scripts, what are you doing with them and what are the symptoms you have observed? Are the problem scripts all on the *same* server or different machines? Are you having problems across multiple hosts or just one host? What did they mean by 'server time issue'?

If the only site still standing is on a dedicated server, would the others happen to be on flaky shared webhosting (ie. hopelessly overloaded) machines? If so, the solution might be to move the affected sites to a less crowded machine (maybe get your own quality VPS). Usually its the database server that falls over first on shared webhosting, so it might be worth looking to see if the scripts can be made more efficient (ie. less database calls).

But that's just a couple of guesses made with not a lot of details to go on.

[newsite12]

Exactly what are these scripts, what are you doing with them and what are the symptoms you have observed? Are the problem scripts all on the *same* server or different machines? Are you having problems across multiple hosts or just one host? What did they mean by 'server time issue'?

If the only site still standing is on a dedicated server, would the others happen to be on flaky shared webhosting (ie. hopelessly overloaded) machines? If so, the solution might be to move the affected sites to a less crowded machine (maybe get your own quality VPS). Usually its the database server that falls over first on shared webhosting, so it might be worth looking to see if the scripts can be made more efficient (ie. less database calls).

But that's just a couple of guesses made with not a lot of details to go on.

Like i said above in my op. I have on multiple servers all on VPS and not the cheap ones... these are good quality 2GB RAM setups with CentOS.

The penny auction by its nature relies upon many db calls. The query setup is trimmed right down to the bare minimum of calls to satisfy the functions.

winwinpluz.com will give you an idea on how it works. If you open an auction by clicking on the product image, you will see the bidbutlers bidding away on the right hand side, also the timer counting down and the time increasing with each bid. Obviously the more connections going to the site, the harder the whole thing needs to work, so this would never function correctly on shared setup. It also has a cronjob running every second, so quite a lot going on.

Like I said.... the websites all ran perfectly for 4 months across different VPS accounts all based in Thailand. Then overnight, they all developed this strange anomaly... sometimes a listing will run fine for 2 days, and other times it will simply count down to zero and end only after an hour or so.

I have also identified hotspots.... just after midnight, and around 09.30 am seems to be the most common times for the problems to occur so maybe it is the hosts running all the automated backups and causing server lag in the db server.... This is one possibility that ran through my mind. We have had many auctions die at 00.02.