| Archive for myfreeforum.org Before posting please check the "stickies" in the support forums. Please ask questions in real English and not "txt". You will get a better response. Please do not ask support questions via PMs. |
| CodyT07 |
Ouch for PhpBB Sitehttp://www.phpbb.com/Aren't the MyFF servers using the same RAID array as PhpBB server? |
||||
| admin (no pm's please) |
Re: Ouch for PhpBB Site
Ouch 
We have had a single disk failure here, and minor engineer idiocy aside got through with a live replacement and rebuild without any down time, which is what you expect from a RAID5 array. To quote wikpedia:
So you have to ask yourself the questions: 1) How slow were they with the replacement? Once you are on one disk failure, the clock is ticking. 2) How many drives involved? You can have up to 8, we stick on 5 as a compromise between reliability and performance. 3) How old was the system? 3-4 years is about the sane limit for reliable operation. 4) Why are they down for days? Possibly an unfair question. If the unthinkable happened, we have spare capacity on servers to reload backups to alternative servers a lot quicker than that. But then again it would be hardly done by the time the server in question was ready to have forums restored to it again, so it would be a tricky decision. It may go to prove that nothing is 100% safe, but the above questions have good answers and bad answers, and we have good answers here! |
||||
| CodyT07 |
Re: Ouch for PhpBB Site
I can answer number 3 3)The release of PhpBB1 was December 9, 2000 So they have to rebuild the 'hardware' part of the server? Out of curiostity (and to keep my learn something new every day trend going) what all goes into 'server maintinence'? I see that happen a lot but never understood what it is about. |
||||
| admin (no pm's please) |
What does the release date of phpbb1 have to do with the age of a server? | ||||
| admin (no pm's please) |
In principle when a single drive in a 5 drive raid array goes and is replaced within 24 hours, the risk of a second and therefore fatal failure is approx 1/10,000.
e.g. statistically insignificant, but this theory ignores factors such as a problem with a batch of drives or if drives are too old, or if some environmental circumstance is effecting a server. e.g. you really cannot rely on the odds being as good as 1/10,000. |
||||
| admin (no pm's please) |
Interesting article here from googles research:
http://storagemojo.com/?p=378 Google use consumer hard drives with a MTBF (Mean Time Between Failure) of 300,000 hours, our drives on the other hand have typically 1,400,000 hours rating. Google find that the theoretical failure rate of 1.46% a year, is more like 6% in the field  That would be quite scary if it did apply to server specced drives. But even then would translate to a theoretical 1/1250 of array failure given 24 hour replacement. But that does ignore the fact that failures do tend to be highly correlated.
On our drives the theory translates to 1/14,000 risk in a year, but if what google has found for consumer drives applies equally to server drives, you could place that at 1/3000 approx. But given we do have offsite backups, such stats are way too small to be concerned about. |
||||
| The Alchemist |
So this is why I can't get the skins from phpbb.com. I thought the site was hacked. | ||||