Cannot access Sammyboy late at nite

[Qantas]

Always seem to have server error late at night. Any bros got this problem? Or issit PAP IB already started their hanky panky actions against Sammyboy like what they are doing to TR?

 

[youtalkcock]

]I have several ideas for you to consider:

1. Add a remote SQL server paying for only 2 months at the most. Shift SQL there and back after GE.
2. Off load SQL db temporarily, make a SQL backup now, purge away historic data keeping only the last 2 month's data, and lighten the load, merge back to backup after GE.
3. Add a static mirror, for reading only - can not post, read only, different URL, based in SG, it clones from your real SQL db. divert main traffic here, and only when users hit POST or anything that writes SQL db, that they will be directed to actual server. That mean a kind of quasi reverse-proxy to take the load / hits for READING users. This way lots of readers can read without SQL db errors.

[/FONT]

Uncle Yap's suggestion for SQL database errors. You will get use to it.

 

[Qantas]

Always seem to have server error late at night. Any bros got this problem? Or issit PAP IB already started their hanky panky actions against Sammyboy like what they are doing to TR?

 

[myfoot123]

Always seem to have server error late at night. Any bros got this problem? Or issit PAP IB already started their hanky panky actions against Sammyboy like what they are doing to TR?

Netizens will literarlly go to the street and support the opposition parties if PAP tried to play blocking out games. They have yet to taste the power of mass outcry in the street, especially now that election date is set. Such risk, they cannot bear.

 

[Leongsam]

I've added another 4 gb of memory as the swap file was too large for the existing RAM to cope with.

Should improve things considerably. Let me know how things go tonight.

 

[uncleyap]

I've added another 4 gb of memory as the swap file was too large for the existing RAM to cope with.

Should improve things considerably. Let me know how things go tonight.

Very logical step Sam! Bravo I think this would be effective.

So an hour ago I tried to access and found that even your Apache was gone, I confirmed by connecting via USA proxy, I tot your server up lorry already . So it was being added with RAM. Well done! RAM is necessary. I assume your server takes ECC DDR3? They are also not that steep priced any more. Kingston non-ECC value-RAM 4GB is @S$53 approx now. I trust this upgrade will tie us through the GE will less MYSQL down time.

 

[Leongsam]

The server has plenty of grunt. It's an E8400 Intel Core 2 Duo 3 GHz machine. The processor load is well withing limits. The problem thus far has been the size of the swap file. The extra memory should help matters considerably.

Very logical step Sam! Bravo I think this would be effective.

So an hour ago I tried to access and found that even your Apache was gone, I confirmed by connecting via USA proxy, I tot your server up lorry already . So it was being added with RAM. Well done! RAM is necessary. I assume your server takes ECC DDR3? They are also not that steep priced any more. Kingston non-ECC value-RAM 4GB is @S$53 approx now. I trust this upgrade will tie us through the GE will less MYSQL down time.

 

[leoman]

testing....testing!!

ok ....all system go

thx

The server has plenty of grunt. It's an E8400 Intel Core 2 Duo 3 GHz machine. The processor load is well withing limits. The problem thus far has been the size of the swap file. The extra memory should help matters considerably.

 

[Leongsam]

There's no point testing now. It's the peak hour traffic that brings things to a standstill.

testing....testing!!

ok ....all system go

thx

 

[leoman]

ok lang, now can go in easy .... coz got put ky har hehehhe

There's no point testing now. It's the peak hour traffic that brings things to a standstill.

 

[uncleyap]

The server has plenty of grunt. It's an E8400 Intel Core 2 Duo 3 GHz machine. The processor load is well withing limits. The problem thus far has been the size of the swap file. The extra memory should help matters considerably.

Yeah I remember from our previous discussion long ago, on the SQL issues, at that time RAM prices were steep, you said it looked like load will be OK if SQL & Apache gets hosted in separate servers, we knew that's going to burn money too severely. Today's RAM is significantly down. RAM really get breakthrough, if you already seen the swap size swell. It is THE issue for sure.

I recently have a 16GB RAM system to crunch all these youtube.com videos to screw LEEgime, I monitored and even with 3 virtual machines running it is not swapping. An alternative use of these RAM is to mount them as RAMDISK, copy entire virtual machine into RAMDISK and they are really performing fast.

I use a shell script to do mount tmpfs & copying.

Sam I think you don't have to really wait for users' peak hours to test the effects, looking at your swap level while you run some heavy SQL loading maintenance scripts will do it!

Cheers!

 

[Leongsam]

Passed the first hurdle which is the mid morning peak... let's see how it goes tonight.

 

[Cestbon]

Maybe U should try to add more memory solve it . I sure more will visit this forum from next week.

 

[Qantas]

OK so far so good. Things improved tremendously tonite. Bravo Sam.