- Joined
- May 18, 1997
- Messages
- 55,664
Actually we were exposed and did not have backups of the database when it cratered this time. We are in middle of big move to FreeBSD. Shit happens. We are working to get this exposure corrected. The three webservers that handle our forums have been moved to FreeBSD, but the forum is still on Redhat 9. We have been afraid to move it due to unknown complications and quite frankly we have some other priorities that pertain to the HardOCP site.Nasty_Savage said:Database Management and hardware Jedi are two different things. If I recall, the [H] servers are fairly hefty spec wise. I think the baby bs going on here is retarded...this isn't a home box /hobby forum. Its not like you have to pay for it (and paying for General Mayhem is less then nothing). Now that its back and my "shakes" are going away, I can resume goofing off at work. Thanks Kyle/Cliff and all the rest.
I am interested in knowing if this came about from an upgrade to Redhat? I thought I read that on the front page. Hopefully the experience can be relayed in a post for us interested nerds
I think the hardware we have is very robust and when the forum cratered, we were not under load. I was using Vbulletin tools to correct tables and optimize them inside Vbulletin and I had turned the forum OFF.
The database is in a single 4U box that has about 125 CFM of airflow through it. Dual 2800 Athlon XPs, 4GB of Ram, with three RAID arrays. All disks are 15K RPM U380SCSI drives 36GB in size.
3 Drive RAID 5 Array for OS.
3 Drive RAID 0 Array for Tables.
2 Drive RAID 0 Array for Index.
OS is on its own LSI SCSI card and the Tables and Index share a 2 Channel card (all 64-bit).
Watching loads and such, we do not think we have any hardware issues.
IN FACT AND THIS IS THE STRANGE THING, we have not seen a database crash with the servers under load since earlier this year. The MySQL database corrupts tables usually when there is hardly any use going on. (Like it did this time.) If you can help us figure this out we are all ears. The HardForums have been working GREAT under load as we generally now have a three to six hour period a day where over 1000 people are on at once.....