Carl Hutzler’s Blog

So even as I published my last post on FreeNAS and how to backup your data to an off-site location I started seeing some errors in my FreeNAS logs. Eventually the errors got worse and worse and then the whole operating system started dying and the FreeNAS box was rebooting itself and failing during backups.

The errors were disk related. Failures to write. Failures to read. Everytime the system crashed and rebooted, it would run FSCK and eventually things would check out. But these errors were at a minimum causing the RSYNC backups to fail regularly.

I initially thought it was the software RAID provided with FreeNAS so I switched from JBOD to RAID0 (stripe). The RAID0 seemed better for a bit, but then it too had issues.

So I bought a RAID SATA PCI card online. It arrives next week. I thought that might be a possible solution as maybe the FreeNAS software RAID solution was not as reliable as it should be.

To hedge my bet on the HW RAID, I also decided to buy two new 1TB drives. This time I would buy Seagate drives. I had been using Western Digital 1TB drives and something in the back of my mind reminded me that the Infrant.com site mostly recommended Seagate drives for their NAS units. So I bought the Seagate ST31000340NS Barracuda ES.2 SATA 1000GB (32MB Cache) that they listed.

Turns out the drives arrived first. So I installed them and set-up my software RAID in a JBOD configuration. I started the RSYNC backups again and they have run flawlessly for a day and a half now (its a lot of data). And the speed of the backups is going even faster now….not the throughput over the network, but the overall wall clock time is noticeably shorter. I am guessing that the errors I was seeing were actually when a read or write had been retried multiple times and eventually failed permanently. But that there were many more errors which were slowing down the transfers using the Western Digital drives. Now that I was using Seagate and things were happier, backups are completing much faster (maybe 2-3 times). It’s not the WD drives were defective….I just think there is something in their firmware which causes issues in RAID configurations…and if you read the Infrant site page, it seems that several manufacturers have problems in this area.

So the moral of the story? Not all hard drives are created equal when it comes to RAID storage (even just JBOD if you call that RAID). I don’t know why the WD drives had issues and why the ST drives work well. But even the Infrant guys seem to say that ST works well and very few other manufacturers work properly. I bet most of the compatibility issues are the firmware on the drives. Wonder why these manufacturers don’t figure out their issues? I guess until then, Seagate can go on charging a 10-15% premium for their drives….I am happy to pay it :-)