Another Dreamhost Disaster

Another suck-ass Dreamhost moment. Their database server went down (like everything here’s been going down recently). They say they backup and everything will be restored, but they don’t backup right or if they do, their restore is worthless.

An entire day’s worth of phpBB posts, PMs, registrations, all gone. The 9th has ceased to exist. It jumps from the 8th to the 10th.

I realize you get what you pay for. I just want to bitch because it’s getting to be where you don’t even get what you pay for. (The trouble ticket is still open and not replied to). <-- board that’s affected

See my blog at or
my personal webpage at

I would like to point out my tickets have been open over 17 hours without a reply. That’s certainly a new record.

See my blog at or
my personal webpage at

It might be a record for you, but it certainly isn’t a record for DreamHost. Relax, dude. DreamHost usually handle support requests within 24 hours. The support queue is quite long at the moment, no doubt due to some recent database issues.

Simon Jessey | Keystone Websites
Save $97 on yearly plans with promo code [color=#CC0000]SCJESSEY97[/color]

Everything has not been ‘going down’ recently. I have very detailed statistics on server reboots, mail performance, mysql performance, etc. Stability and performance have been steadily improving for months now.

The data on the mysql servers is backed up nightly and the live data is stored on a raid 5 disk array. In this case two drives failed at the same time, which meant the raid did not have time to rebuild a parity drive on a spare. Dual drive failure is very rare, but the nightly backups are the safety net for such an occurrence. In this type of case you will lose the data between the last nightly backup and the dual drive failure. Still the system worked as designed.

Any data loss is painful and I have credited your account for one month of service. Our customers have been very clear that they want the best hosting possible that costs about $10 a month. Currently, fully mirrored database machines each with their own raid 5 and hot live backups is not feasible at that price point, but we are always on top of the server component industry prices and are quick to upgrade our plans to reflect the current market.

I should point out that this was a single database machine (there are hundreds.) It was a single database machine, dual drive failure and not a widesread database system failure. Ordinary drive failures are not related to the network or other machines in the datacenter.