Read an article over at LH earlier this week that convinced me to buy some more backup storage before the price hikes rippled through the external storage market like they already had for the internal HDD market. What? you ask? The price for internal drives has change from $65 to $129 for a slow 1.5TB 3.5" disk since the flooding in Thailand. Some of the non-tech specific retailers haven’t changed their prices to reflect the coming HDD shortage, so getting some storage to fill your needs for the next 6-12 months might be a good idea.
I’d been looking for storage for about a month. I really wanted something like a 4 disk eSATA array, but with the HDD prices doubling, that is out of my price range now. After I read that article, I knew some sort of solution would be needed. Costco, Target, Walmart, OfficeDepot, Staples were all checked. The local Target seemed to have the best price for a 2TB external HDD – $90, but they didn’t post the warranty information on their target.com site. That store could be on my way for the weekly grocery trip. In the store, someone askes if I’d like some help. BIG MISTAKE. Seems he’s the cell phone guy, but he knew where the HDDs were and started into his upgrade your cell phone sales pitch. That’s a waste on me.
Staring at the 2TB Seagate Ext … only a 1 yr warranty. 6 inches away is a WD MyBook 2TB for $100 with a 2 yr warranty. Is an extra yr warranty worth $10 to me? YES! I know HDDs that fail in the first few days of heavy use, but they also fail about a year into normal use. Longer warranty is good. Of course I’ll also use the Amex CC to double the warranty – ok, Amex only adds 1 more year for warranties over a year, still, that’s a 3 yr warranty. Good enough. Bought.
I really didn’t want a USB HDD at all. I’d rather have an internal HDD for speed and since it will get cooled and I wouldn’t have to worry about a flaky USB connection. To start, I’d drop an internal drive into an already connected dual dock which has been working well enough.
Am I the only person that has flaky USB connections?
I break down and connect the new HDD via USB to the storage server. I wipe the crapware included on the disk – MyBook crap. Next, repartition and make a file system. I force an extended partition and create 1 large – include everything partition in the first logical. This is probably a bad idea for future flexibility, but I really want to force myself to use this disk for backups and only backups.
$ sudo fdisk /dev/sdh
$ sudo mkfs -t ext4 /dev/sdh5
I mount umount the old disk and put the new disk as /backups. Then I create per-machine directories to keep the backup locations organized.
$ sudo mount -t ext4 /dev/sdh5 /backups
$ cd /backups
$ sudo mkdir regulus romulus lubuntu xen51 dms44 pki52 crm46 moms ….
You get the idea.
Next, I need to move a rather large, 331GB, virtual disk from another HDD to help move some other files around so that virtual disk can be loaded onto an internal HDD for better performance. Lots of temporary moves are planned for the next few days. These files will exercise the HDD, get it plenty hot and hopefully see any crashes before any important data is used. I kick off the file copy and wait.
Gee, then is taking much longer than I thought. I check the write speeds – around 31Mbps. Damn USB2! This is gonna be awhile. Then it stops at around 135GB. I switch from cp to rsync and try again. 135GB and it stopped. umount/mount and try again. 135GB and it stopped again.
While this is happening, I update the fstab with UUIDs. It is nice to actually control where devices get mounted every time.
I need to do a little rsync research. Seems that rsync needs a temporary place to write some data and the default location isn’t big enough. Rerun rsync with -T option. 135GB and it fails. Further, the disk seems to have gone off line. dmesg shows all sorts of errors. hub_port_status failed (err = -110) and others. What’s worse is the entire system has become slow as the kernel tries to resolve the USB issues. Not good. No other hardware on this system has changed in a few months.
I surmize the USB extension cable that reaches around the back of the server could be the issue. It hasn’t been an issue before, but … perhaps. Switch to using the front panel USB. Try again. This time it is really bad.
USB still sucks
Sure, for a quick way to connect a disk and move files it is fine, but for heavy use or long term mounts, forget it. eSATA, SATA, or even NFS are better. So here I am with a slow system, unresponsive disks. umount, mount aren’t working either. Every attempt at those commands lock up. Unresponsive. The only way I know to deal with this is to reboot the system. That has to wait.
The reboot waits until the following morning. Seems the things that I didn’t want rebooted had locked up overnight. Thanks USB. This server has been running without any issues for years. It has been running these VMs for months with out issue. The only errors in syslog are related to USB crap. Yes, USB sucks.
At reboot, the USB disks aren’t recognized. Unplug them. Reboot. Tell the OS to skip mounting the missing USB disks. Everything works as expected. I connect the older USB dual dock and mount it. Nice. Next I connect the new MyBook and it fails to be seen by the OS. My first thoughts are that the enclosure is defective, so I should return it to Target for another. I’ll ponder that a little more. In the meantime, I connect it to a WD TV Live HD media player with a tiny bit of hope … but that doesn’t work. It sees the drive, but since it isn’t NTFS for FAT-whatever, the video player doesn’t know what to do. Guess I expect too much.
At this point, I’m stuck with a 3 day old 2TB HDD that really can’t be used on any of my systems. Having a Linux-based file system is critical to me. Backups to NTFS aren’t really an option. The file permissions need to be maintained or the backups are nearly worthless.
I’m loath to pull the HDD out of the enclosure this soon. The storage server can’t handle anymore HDDs and there isn’t any more physical room for a PCI or PCIe adapter inside that machine. There is an empty slot, which makes this really frustrating.
Reading some reviews found searching for 2TB mybook linux usb3 and it appears there’s a major firmware bug likely. Time to return it.
The best answer I have today is to migrate the running VMs from an ESXi machine here, then repurpose that to be a KVM/LXC server going forward and add the new backup storage. Unfortunately, my timeframe doesn’t support doing all that. It appears my quick solution will need to ignore the 2TB HDD.
I wonder if buying the Seagate would have turned out better?