- Joined
- May 18, 1997
- Messages
- 55,602
I has been a long 6 months or so......
We think we have finally found the problem with MySQL that is causing us issues. We hope to have a couple of options to fix the problem tomorrow and we hope the solution in place by end of next week. This means we should be up and running uber-freaking fast again and back to being stable like were were a year ago.
The problem seems to be a "compatibility" issue between 64-bit Linux, MySQL, and 8-way processors. And the fact of the matter is that when we configured this, there were very few people running dual 4-core processors to rely on for feedback experience and we were told by more than one expert that our configuration was good to go. I will tell you that we did a solid month's worth of research into this and at that time there did not seem to be any problems. Also, we used the configuration for a full 30 days in testing and then were in production a full 30 days before we ever saw an issue. This massive upgrade was not something we rushed into (for a change).
It has taken us about 4 months to figure out what is wrong. Now that we are 95% sure of the issue (which is random MySQL crashes) we should be able to put a plan of action in place to fix everything. At that point we will be back on track and we think the $15,000 worth of hardware (2 new mult-processor database servers as well as a new multi-cpu webserver addition (total of 3 webservers when we are done)) that has been purchased through this upgrade will really shine.
Please be a bit more patient with us, we should have the search server turned back on next week and we will be running at full capacity. At that point, if you have slow page loads, you can most likely blame it on your ISP instead of use for a change! I do appreciate you guys being patient and hanging in there with us. Thank you.
We think we have finally found the problem with MySQL that is causing us issues. We hope to have a couple of options to fix the problem tomorrow and we hope the solution in place by end of next week. This means we should be up and running uber-freaking fast again and back to being stable like were were a year ago.
The problem seems to be a "compatibility" issue between 64-bit Linux, MySQL, and 8-way processors. And the fact of the matter is that when we configured this, there were very few people running dual 4-core processors to rely on for feedback experience and we were told by more than one expert that our configuration was good to go. I will tell you that we did a solid month's worth of research into this and at that time there did not seem to be any problems. Also, we used the configuration for a full 30 days in testing and then were in production a full 30 days before we ever saw an issue. This massive upgrade was not something we rushed into (for a change).
It has taken us about 4 months to figure out what is wrong. Now that we are 95% sure of the issue (which is random MySQL crashes) we should be able to put a plan of action in place to fix everything. At that point we will be back on track and we think the $15,000 worth of hardware (2 new mult-processor database servers as well as a new multi-cpu webserver addition (total of 3 webservers when we are done)) that has been purchased through this upgrade will really shine.
Please be a bit more patient with us, we should have the search server turned back on next week and we will be running at full capacity. At that point, if you have slow page loads, you can most likely blame it on your ISP instead of use for a change! I do appreciate you guys being patient and hanging in there with us. Thank you.