FreeNAS - Anyone Tried it as a Music Server?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The way I read it was that the drive might not respond at all, not even with an error message, for more than 8 seconds. Once that occurs, the controller drops it from the array. Then the array must be rebuilt, which can take days. During the rebuild, the array is vulnerable such that if the same error occurs again, all the data is lost. My interpretation was that if the drive at least returns an error message, the array is put into a temporarily bad state from which it recovers quickly without needing a rebuild.

Yes, I agree, however it seems that what the WD paper is saying is that the 8 sec timeout is often a false failure - so replacing the drive and rebuilding the array is not really required. That is, assuming the drive was able to replace the defective sector with a spare. Ah, but there is possibly a hole in the data (assuming retries did not recover all the original data) so the data does need to be scrubbed, but no hardware needs to be replaced, assuming the drive is not on a rapid decline.

Just thinking out loud here. They suggest acking quickly to the controller, it did not provide good data (Edit: as I understand it, perhaps it does through ECC) so the RAID parity (when used) solves that problem? It puts off doing the repair as a background task - but if it repairs it with a spare sector is it able to recover the data? Not sure how the self repair works, the only way this makes sense is if the drive swaps spares in on detecting a correctable (internal) error so that it still has good data, or perhaps a threshold where re-reads with error correction gets the good data so that the spare block has valid data. No RAID scrubbing would be needed if this were the case. I'm just not sure exactly how these self repairing drives work.

Seems we have 3 possible cases here, a spare with good data is put in place, a spare with bad data but still a "good" working drive. Or the drive is out of spares and actually needs to be replaced. I suppose if the deferred repair does not work, the drive will take itself off line and be noticed on the next access.

Seems to me that if the WD paper is correct, then all that's needed with desktop drives is to clear the fault and scrub the array. If the scrub is able to clear the fault then the system is back to a good state, if not the drive needs to be replaced.

BTW, got to wonder what could possibly take 8 seconds, that's a long time - even for a disk drive!
 
Last edited:
Back to FreeNAS, I have a P4 tower here and I added a SATA interface card, and large (1TB) hard drive. Burnt the ISO to CD now just have to read the manual as to how to set up FreeNAS. Also added a USB 2.0 card for backups, and with an internal connector where I might use a small flash drive for the OS.

Also made an Ubuntu live CD but it would not boot, not sure why or if the CD image might be bad.
 
Last edited:
Dunno. The internals of RAID controllers are way beyond my knowledge and experience, so I can only speculate. As a user, at least I know that if I use a desktop drive with hardware RAID, I should run a utility that configures the drive to return an error message when it gets into a self-repair state, making it in effect act like an enterprise drive.

To look at my array, I can go into the Promise WebPAM and examine each drive using the "BSL" (bad sector log) feature. I don't have any bad sectors on any of my drives, fortunately. I have good airflow in my case, with a fan pulling air across the drives in the array.
 
With all the variations and enhancements to the IDE, ATA, and EIDE standards it is hard to keep track of it all, and I'm not sure that the companies publish the details that they might consider trade secrets.

Found this for anyone who might be interested; I'd sure like to be able to read out all that data. Have to look around for some software:
http://en.wikipedia.org/wiki/S.M.A.R.T.

Interesting that it is stated there that SMART has not proved accurate in predicting failures, but that " . . . – specifically, in the 60 days following the first scan error on a drive, the drive is, on average, 39 times more likely to fail than it would have been had no such error occurred."

My intuitive feel was that if the drive is taking more than 8 seconds to correct, something real serious is wrong with it.

Edit: Smartmontools: http://prefetch.net/articles/diskdrives.smart.html
Download: http://sourceforge.net/apps/trac/smartmontools/wiki/Download
 
Last edited:
Interesting how well the Intel ICH10R Southbridge does here in RAID benchmarks:
http://www.tomshardware.com/reviews/ich10r-sb750-780a,2374-10.html

However, they used solid state drives, which while providing impressive big numbers are far less common than the usual disk drive. They do not have seek time or rotation latency characteristics so if any of the designs in this benchmark have features to improve performance in these critical areas they will not show up.
Interesting that the Intel chip in particular does so well.

My two best systems have the ICH7R which also supports RAID.

Pete B.
 
I've been reading about FreeNAS and it seems to be a good solution for network storage:.[/QUOTE

It's just BSD Unix. It is well proven being 25+ years old. Also seeing as BSD is what's inside Mac OS X it is also the second most commonly used OS after Windows.

There are many ways to re-package and distribute BSD and all routes basically get you the same level of performance and reliability.

As for using a USB memory stick for primary storage. The USB interface itself will be a bottleneck. Even if the flash ram was infinitely fast it would still be on a USB port which can't match the performance of the SATA conection, even if USB has a higher raw bits per second spec.
 
I'm aware of the Unix history, just wondering how well FreeNAS is packaged and if there are any issues.

USB? You're talking about my other thread? I don't expect to do much better than a single SATA 3.0 drive on raw throughput, unless I can stripe 5 or 10 USBs, however the mechanical latencies, seek time and rotational, are what really hurt a disk drive's performance. These should be minimal, but are not as seen in some test reports, for USB Flash drives.
 
I'm aware of the Unix history, just wondering how well FreeNAS is packaged and if there are any issues.

............

It seems to be packaged quite well. I know nothing about BSD and I had it up and running in no time. I haven't messed with it since I built it.

Trial it on a junkbox real quick, won't take you long.

I don't know.... does the speed of the drive that the OS is on affect more than boot time?
 
It seems to be packaged quite well. I know nothing about BSD and I had it up and running in no time. I haven't messed with it since I built it.

Trial it on a junkbox real quick, won't take you long.

I don't know.... does the speed of the drive that the OS is on affect more than boot time?

This is exactly what I was wondering if it was reliable and easy to maintain. You're experience is good input thanks.

I'm thinking of the SSD drive for the other PCs connected to the NAS box. Planning to boot from CD, and use flash or a floppy for the configuration data on the NAS.

As far as the other boxes go, any OS application that is on disk will load faster from a good SSD. Office apps, multimedia or music player program etc.
 
I'm thinking of the SSD drive for the other PCs connected to the NAS box.

One place I worked did not install any kind of disk in the desktop machines. Each of the desktop computers booted from the network. When a user logged in his home directory was mounted from the server. This meant that
1) If a computer broke it could replaced with a new one instantly no disk means zero configuration of desktop computers as they hold zero local data. And YES a gigabit Ethernet is as fast as a local SATA disk, or faster.
2) Any user can use any PC. If PCs don't hold local data then it matters not which desk you set at. Any PC will display your desktop and files when you log in.

Planning to boot from CD, and use flash or a floppy for the configuration data on the NAS.

Why? That is just plain un-needed complexity. Obviously a NAS server has a disk. Keep the OS and the config data on the disk. That what it's for. Also this will get backed up up when the backup script runs.

The bottom line is that if you have large and fast central storage and a fast enough network you do not need any more storage
 
I'm fully aware of how many businesses set up their networks. My current client does not have gigabit ethernet, so we boot locally but have user data come up from the server so that we _can_ go to any machine and login. Pretty standard and obvious.

The fact is that we do not have wired Gigabit ethernet in our home, in fact several of the machines only have 100 MB ethernet adapters, or are wireless.

As I said earlier, I'm not looking to build a high bandwidth server, just enough to serve audio and video media. This is more cost effective since all our systems have at least a small boot drive right now.

The configuration that I mentioned, booting from the CD with config data on a floppy or USB stick is suggested in the FreeNAS literature. It also means that you can boot even if you're having drive problems. I suggest that you read the literature, or even watch the video at the start of this thread.

Pete B.
 
On the other end of the complexity scale. iTunes will do everything most people need. Any computer running iTunes can makes it's entire library available to all other computers on the network. The simplest media server then is just a Drobo raid box setting on top of an older G4 powered Mac Mini. The entire setup uses so ittle power that you can afford to let is run 24/7

That is a consideration. Power is not cheap. A full up PC chassis can draw 200 watts. At 20 cents per KWH it adds up to $28 per month or $345 per year to run the PC. A used Apple G4 Mini will pay for itself if it replaces a standard tower pc chassis. For those who don't like Apple's version of BSD Unix, you can run other OSes on same hardware. The G4 mini is basically a small, low powered notebook but without an LCD screen or keyboard but with a Firewire port for high performance disks, a perfect home server platform. At one time I did run a PC based server but then I looked carefully at the cost of the power to run it, replacing it was a quick payback. A PC stuffed with disks really can cost $30 a month to run. That's about $1,000 over the life of the server. Double those numbers if the server runs in an air conditioned space.
 
Last edited:
It is working:
Dell Precision 340, 2.2 GHz P4 FSB:400, 512 MB ECC Rambus RAM.
Parts added from NewEgg were:
PCI to 4 port SATA-1 card MASSCOOL XWT-RC040 RT $10.99 - open box.
1TB Seagate 7K ST31000528AS OEM $73.99 - free shipping.

The only way the 1TB drive will saturate the SATA-1 interface is during a cache hit, and since the point is to stream data at a rate much lower than SATA-1 - it is not an issue.

Booting from a Live CD, floppy for config storage. Works fine. FreeNAS updates are just burning a new ISO - nice.
Take the CD out and it boots XP from the small ATA-100 drive.

It does 11.2 MB/sec on a drag and drop of a 3.7 GB folder, over 100 Mb/sec wired LAN. Both ends are Intel Pro 100.
Gigabit Ethernet would be nice for uploads ...

Difficult to stopwatch time the time from double click on a music file to music playing in Windows Media player. About 1 sec, and it feels real fast.

Pete B.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.