Important Downtime information between 14th and 25th of October
On 14th of October my old webserver’s hard drive died of old age, bringing the entire machine to its knees. An ext3 filesystem corruption has also occured as a side result, taking part of the mysql datafiles with the dying drive.
Me and my coleagues (who I share this server with) have decided to buy a new webserver. So we bought this little pizza-box (a refurbished IBM eServer x345):
The machine’s core consists of two Intel Xeon 3.2 processors, backed up by 3 gigs of RAM, powered by two redundant PSUs. Data is stored on five 10kRPM SCSI disks, 36GB each. RAID5 makes sure that even if one of the disks should fail, all the data is still safe. Two Gigabit Ethernet ports are connected to the datacenter’s backbone. We are planning to buy some spare hard drives, so the array will be eventually extended with sixth hot-spare hdd, which role is to take over immediately at the time of failure. Such redundancy allows the machine to continue service even if two hard drives should fail.
Unfortunately I was very busy during the last week, so I could not immediately bring my website back. After 11 days we have finally brought the box to the datacenter. It has been online for the past 6 hours and I have already recovered most of the broken datafiles.
What happened is actually very positive. The old server should have been replaced long ago, and this box, even though it’ not entirely new, has four times more juice than the old one. I am not going to mention the redundancy of its core elements.