Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Delta-Wye
Sep 29, 2005
I don't know if you guys have seen the udat. It's what I would like to do. A decent computer plus a few drives in a small form factor sounds awesome to me. Maybe not quite so small - I wouldn't mind putting a 'real' computer in there with a gig ethernet port and maybe more muscle, but the idea is certainly there, and the form factor is awesome. Mashie does pretty nice cases, the Anemone is one of my all-time favorites.

Adbot
ADBOT LOVES YOU

Delta-Wye
Sep 29, 2005
I've been thinking about rolling my own NAS. I like the idea of installing FreeBSD and putting it together with zfs and all the other good poo poo. Orignially I was looking at some of the fancy 3.5-2.5 drive bay adapters and was considering rolling my own case - I'd like to have something small and unobtrusive. I found a pretty good mini-atx case though, the Chenbro ES34069 Mini-ITX Home Server/NAS Chassis, which looks like its made exactly for my needs. Only two downsides - I don't have a mini-itx board, and the case is pretty spendy. The JetWay JC62K AM2+/AM2 NVIDIA GeForce 8200 Mini ITX AMD Motherboard from Newegg looks like a good pick, assuming it fits adequately. I have 1) several sata drives that will work well, and 2) an extra laptop drive to use as a boot drive.

When I add the mobo, case, ram, slimdvd drive (not necessary, but would be cool to have it I think), it comes to ~$400. Is there a better option or solution to my goal? I think this would result in a pretty awesome product. It also has gigabit ports, which is good for speedy data transfer, although I'd have to upgrade my switch to get full use out of it. It has TWO, so I could even use this box as a router/firewall down the road (although NAS+firewall is kinda a lovely combo). What do you guys think? I wouldn't be surprised if there are better motherboards out there - Newegg seems to have a pretty small selection.

Delta-Wye
Sep 29, 2005
I think I'm going to go ahead and build my NAS box - however, a good portion of the data that is going to be stored on it eventually is already on a pair of the drives I want to use, and honestly, I don't have the extra storage space to hold it while I build the box. I want to use raidz, how badly is it going to mess me up to use two disks now, and add two later?

According to wikipedia: It is not possible to add a disk to a RAID-Z or RAID-Z2 vdev. This feature appears very difficult to implement. You can however create a new RAIDZ vdev and add it to the zpool.

Will the two vdev's mount as two seperate devices? I hate having multiple drives to deal with and would prefer to have everything as a monolithic device. To tell the truth I'm not real familiar with the Zfs lingo yet, so I thought I'd ask before I threw down some money for the hardware.

Delta-Wye
Sep 29, 2005

complex posted:

I'm not sure what you mean by "mount as two separate devices", but ZFS can do what you want.
That's freakin rad. The more I read about ZFS the more I like it. Let's see - I want to make an array of 2x320G drives, and later add an array of 2x250G. I expect to see 570G. Later I'd like to eventually switch over to 4x320G (or bigger). Could I just swap the 250s out one by one and let it rebuild? Would I expect to see 640G or 960G?

Delta-Wye fucked around with this message at 09:54 on Jun 29, 2008

Delta-Wye
Sep 29, 2005
Crazy idea - raid-z guys, back me up (or tell me I'm crazy!)

So, right now I have the following disks:

320G
250G
250G

The 250s are currently holding all my data in a striped configuration (nothing particularly critical, don't need feedback on the stupidness of this arrangement). the 320 is free. I wish to build a 4 drive raidz configuration that will eventually hold all my data. I would like to use these disks + 1 more. My parents want another disk for one of their boxes, so I was figuring here's an opportunity to buy a couple disks, sell them one of the 250s, and then use the 250+320+2 new disks in my new box/array. However, I'm at a loss at how to transfer the data over.

If I can find another disk, I could use the 320, the two new drives, and the temp drive to build the raidz. Transfer the data over, and then remove the temp drive, swapping it for the extra 250. I'm hoping the raid will begin rebuilding onto the 250. My main box has a mirrored 74 gig raptor (rar!) system disk - I could break the raid, and use the disk for a while (I've done this before with no ill effects, when the disk is returned it syncs up and all is well). With a 320g, 750g x 2, and a 74g, I anticipate seeing ~215g. This, plus a few other machines, plus some dvds, ought to be enough to shuffle my data around. Later on, when I need space and have money (aka the stars align) I'd like to swap the 250, and eventually the 320, for bigger 750g drives. I anticipate this giving me ~750 gigs from the start, ~960g after replacing the 320g, and 2.2T after replacing the 320g.

Based on my reading on raidz, all of this should be possible (and relatively easy after I get the knack of it!). Does this jive with your understanding? I hate spending money and making plans when I don't fully understand zfs, and I won't fully understand it until I've spent money on parts and played with it some. :( It is a bit of a chicken and an egg situation, if you understand.

Delta-Wye
Sep 29, 2005

porkface posted:

Which virtualization software would you use for this? I've been trying to do a similar thing with physical disks in VMWare 1.x and it'll only recognize the first 128 GB of each drive.

You ever have one of those moments where you go "drat, thats a good idea... I should have come up with that!"

http://blog.sourcehosting.net/2008/03/18/freebsd-70-vmware-image/ + vmware player = testing.

Now if only I knew the mobo I want to use supported freebsd - something tells me it probably doesn't. :(

Delta-Wye
Sep 29, 2005

Toiletbrush posted:

:words:

Thanks for the info. I've used Linux extensively, and OpenBSD and FreeBSD less so (although I like them a bit better) but the few minutes I sat in front of Solaris made me want to kill myself. Perhaps I ought to suck it up and try again!

I'm going to try and build a Chenbro ES34069 mini-itx NAS, which has room for four SATA drives and 1 2.5" drive (which I already have). The 2.5" will be the system disk.

The only mini-itx board I can find with 4 onboard SATA at newegg is the Jetway JNC62K - beyond the obvious "its a Jetway, duh" I can't find any info involving Linux or Freebsd, let alone Solaris.

Delta-Wye
Sep 29, 2005

Toiletbrush posted:

You've got to be making GBS threads me. Even the normal Solaris 10 package comes with an installer that's not really harder than the FreeBSD one, and it even comes with Gnome.

Try the OpenSolaris 2008.05 package. If that's still bothering you, then I don't know.

For what it's worth, the last time I ran Solaris was... 2001? 2002? Long before OpenSolaris, I'm pretty sure it only ran on Sparc systems then. And it wasn't installing it, it was just using it that I found distasteful. It just bugged me for whatever reason. I'm going to hunt for Solaris vmware image to play with over the weekend I think. It's not that I couldn't or can't figure it out, it's that I don't want to. :colbert:

Delta-Wye fucked around with this message at 22:55 on Jul 3, 2008

Delta-Wye
Sep 29, 2005

Delta-Wye posted:

the Chenbro ES34069 Mini-ITX Home Server/NAS Chassis

Ahem.

Delta-Wye
Sep 29, 2005

Bender posted:

Thanks to some happy circumstances I'll be getting 4 750GB drives, and I figured I'd put together a nice Raid 5 fileserver. Is it a better idea to include the OS partition on the Raid, or to put it on its own, dedicated drive?

I think almost always its better to go for its own, dedicated drive. I'm a bit paranoid, but in the past I've used two small drives mirrored as my system disk.

Delta-Wye
Sep 29, 2005
I'm going to start backing up my webhost to my zfs-based NAS, and i have a crazy idea. Let me know if it's crazy or if I'm missing something.

I want to use a cron script to pull the current web page/database/svn repository via sftp or rsync to a dedicated pool, and take a snapshot every time. This means I will have a historical record (I can roll back to any arbitrary backup if I need to) and the snapshots should be relatively low-cost compared to separate dated backups, especially considering that only a small number of files will change between backups.

Is doing that many snapshots a bad idea? I haven't done much with snapshots on ZFS yet, can you pull out several historical revisions, or once you roll back to you lose all of the later snapshots?

Delta-Wye
Sep 29, 2005

evil_bunnY posted:

Or you just copy them to different directories on a dedupped pool, presto.

I didn't even know ZFS had this feature! A file-level dedup is exactly what I am looking for, and dropping the files into different directories in the script would be trivial. Thanks.

Delta-Wye
Sep 29, 2005

adorai posted:

just remember it's block based and increases RAM utilization by a large amount.

poo poo, I should have continued reading, I read this (https://blogs.oracle.com/bonwick/entry/zfs_dedup):

quote:

What is it?

Deduplication is the process of eliminating duplicate copies of data. Dedup is generally either file-level, block-level, or byte-level. Chunks of data -- files, blocks, or byte ranges -- are checksummed using some hash function that uniquely identifies data with very high probability. :words:

What to dedup: Files, blocks, or bytes?

Data can be deduplicated at the level of files, blocks, or bytes.
But didn't get to:

quote:

ZFS provides block-level deduplication because this is the finest granularity that makes sense for a general-purpose storage system. Block-level dedup also maps naturally to ZFS's 256-bit block checksums, which provide unique block signatures for all blocks in a storage pool as long as the checksum function is cryptographically strong (e.g. SHA256).
and assumed it was configurable to different 'levels'. I would have read more before actually using it, but that is kind of non-ideal for my needs. On the upside, it would be a very small storage pool - I think my site probably only has a hundred megs or so of stuff, and it's mostly going to be completely duplicated from one backupt to the next; I suspect it will be under several hundred megs for a while. It looks like I'll have to crunch some numbers to see if this is a worthwhile proposition.

Delta-Wye
Sep 29, 2005

evil_bunnY posted:

Wait why wouldn't dedup work? It's a trivial amount of data.

It probably would. Alas, my version of ZFS is way out of date and doesn't support dedup.

The cobbler's shoes and all that.

Time to upgrade Solaris/jump ship for FreeBSD/something I guess.

Delta-Wye
Sep 29, 2005

Zorak of Michigan posted:

Dedup would be a lot more CPU and memory intensive than snapshots, so I'd think you'd want to explore snapshots a little more. When you talk about pulling snapshots out, what exactly are you looking to do? I ask because if you just need to copy files from a snapshot to the web host, that's easy and totally non-disruptive, as Fishmanpet says. You just go to the snapshot directory, find what you want, and copy it over. In your particular use case I don't think you'd ever want to actually restore the whole snapshot, so a lot of the situations that I have trouble wrapping my head around would never come up anyway.

I have two situations I'm concerned about :
1) No offense DarkLotus (everything has been great so far!), but my hosting takes a dump and all my data is gone. In this case I want a newish backup I can restore immediately. For this, I just need to grab a copy every week or whatever, and have the newest one available. In this situation I don't even need to keep historical copies.

2) Security breach of some kind that I don't notice for a while. In this case, I need to troll through backups and find the latest copy without the problem, restore and update to fix the initial breach. In this case, I may need to mix-and-match files - the web and database content may be before the breach, but I could need the newest SVN to get my development work back. Having dated backups I can easily poke through and grab files from is good here.

Honestly, I need to do a bunch of upgrades to the NAS including larger disks. If I upgrade the disks, I will probaby have a couple TB of empty space and can easily just keep duplicate copies of the website, storage space be damned. However, I've got this fancy filesystem so if I was going to roll a script to download the contents, doing a snapshot of the directory wouldn't be a huge issue. Upgrading so I have dedup as an option would be cool too; it would be transparent in the sense that I would have a pool full of directories with timestamp names, and they would all look like full backups but the FS layer would not duplicate blocks (which would be a majority of the data).

FYI, I think the box does have a relatively small amount of ram (2GB if memory serves) and I don't know if the microATX board supports more. If the filesystem isn't mounted, I assume there is no penalty for having the dedup option set? It would make sense that ZFS wouldn't keep information about an unmounted filesystem in memory but I'm not sure how it's organized under the hood.

Delta-Wye
Sep 29, 2005
Fellow packrats, if you were buying new drives to upgrade your NAS, would you be willing to go with one of these green drives?

http://www.amazon.com/dp/B004VFJ9MK/

I understand the line between "don't bother replacing, but don't buy new". The price is certainly right, but they are WD green drives. Right now my ZFS-based system has 4 320GB drives and it's time to expand. I think at least a couple are already WD green drives and I haven't seen much in the way of problems with them. At $100 a piece, I can stand to get 4 right now and triple my storage space.

If there is something more suitable I'd be willing to consider it. Those new Red drives look fancy, but I'm sure they'll be more $$$ and I'm not sure if I'll really need the features.

Delta-Wye
Sep 29, 2005

Krakkles posted:

Picked up 4 of these. With coupon code EMCNCJN23, they're $99 a piece, I figured that was a good price.

Now just waiting for a deal on an N40L and WHS2011.

Those are the same drives I'm asking about, at the same price without prime shipping :ssh:

The price does seem pretty damned good though, as long as you don't look too far back:
http://camelcamelcamel.com/Western-Digital-Caviar-Desktop-WD20EARX/product/B004VFJ9MK

Delta-Wye
Sep 29, 2005

evol262 posted:

Sorry for jumping to conclusions, there. Yes, it's 2.0. You can happily put it in a x4 slot.

Welp, bought $400 worth of those 2TB WD Green Drives last night. I realized that going from 4x 320GB would have give me 6x the space too! :cool:

Now I just need to figure out the easiest/best way to go about upgrading the system... decisions, decisions...

Delta-Wye
Sep 29, 2005
I just got in my new set of disks and am trying to decide the best way to go about upgrading from my smaller harddisks to the larger.

The easiest way is to simply swap the disks, one at a time over the course of a couple days until all the rebuilding is done. Viola! More space.

This is hampered a bit by my desire to upgrade the system in general. I would like to reinstall the OS and upgrade the version of ZFS I am using. Right now I'm running some really old version of OpenSolaris and I'm not terribly pleased with it. I would prefer to switch to FreeBSD I think - I've used it before and I find it a lot more comfortable to work in/keep up to date/etc.

To complicate things, the current drives are running some really old version of ZFS - I was asking earlier in this thread about backups and people mentioned deduplication... My version of ZFS must be <v21 because it doesn't support it :( I'm thinking now would be the best time to make a clean break.

I don't really want to lose a bunch of data through experimentation (I'm going to back up the critical personal stuff but there is a lot of other stuff that isn't really worth backing up, but I would also rather not lose) so I'm hoping to kind of bounce a few ideas off of you guys and see what you think.

My machines root file system exists on a harddisk separate from the storage disks which should simplify things. I should be able to export the zfs (pool? i get my zfs lingo confused), install freebsd, and then import the storage harddrives in the new OS. Would I be able to upgrade ZFS versions in place? if that is the case, I could upgrade the filesystem, swap disks, and be good to go I think. If not, could i run the old disks on an old version of ZFS and make a totally new storage set with the new disks and just copy stuff over? I'm not sure if the machine has enough SATA ports for that, but I may be able to get something figured out.

Am I missing some easy method or approach to this? I also need to replace the case because the USPS totally destroyed it (I really liked the Chenbro ES34069 before it got smashed up), and could potentially drop a few more dollars on an overall hardware upgrade. I'm hoping switching to FreeBSD will make it easier to make the server a bit more multipurpose so having a better proc/more ram may not be a totally bad thing. If that ends up being a 'good idea' I could easily just build the new box with the new harddrives and copy everything over across the network. Time consuming, but safest.

Delta-Wye fucked around with this message at 01:30 on Jul 21, 2012

Delta-Wye
Sep 29, 2005

hotsauce posted:

I have an ancient HP Mediasmart EX487 server that I haven't used in over two years. I fired it back up today so I can get my old files (obviously not important at all) off of it so I can get rid of it.

I cannot, for the life of me, connect to it. I connected it to my laptop via ethernet, used remote desktop to log-in and I don't seem to remember the name of the server. Tried Admin, Administrator, and two names I'm positive I assigned to it. No dice. I also installed the Home Server Connector software and it never found it (even when connected to my router). Not once have I found it on any network search on my computer. I tried to find it by typing in random addresses (192.168.1.12, etc) but no dice. I rebooted the server and my laptop a few times in the process just in case. Nothing.

It's as if it doesn't exist.

Anyone have a tip as to how to find this thing on my network/RDC so I can access then wipe?

Do you have access to the dhcp server? if you have a wireless router or something you should be able to login and see if it grabbed a lease. If it's statically assigned, you may have to sniff traffic with wireshark or something to see if it's showing up.

Assuming it's even booting properly and everything.

Delta-Wye
Sep 29, 2005
I did some research and started getting my gear sorted. I'm currently resilvering new drive 3 of 4. I thought I remembered hot swapping the drives previously - pull one out, slap a replacement in, and everything worked automagically from there. No this time! After some aborted attempts, it seems the easiest way is to bring the next drive to be swapped offline, do the hardware swap, do a `zpool replace` with the new drive and wait for the resilvering to finish. iostat -En tells you which logical name corresponds with which serial number so I wrote down which serial number was physically in which position before I turned the machine on.

It looks like I should be able to export the pool, install a new operating system, and import the pool again without any trouble. Upgrading versions for the filesystem seems pretty simple too, and can be done in place.

Everytime I play with zfs I'm a little trepid - I'm inexperienced and I usually learn best by making mistakes and I don't really want to make mistakes with my server. However, with a little reading and careful commands, it usually just works. Want to replace drive with larger capacity ones? Want to upgrade? Want to do whatever? It just works.

Hopefully by this time tomorrow I'll have the last drive installed and should have the larger pool which has me a bit excited! I'll probably give it a couple weeks before dinking around with the OS, but everything so far has been easy.

EDIT: I didn't buy the red drives though :ohdear: Am I going to regret this?

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

I never knew about iostat -En to see serial numnbers, except it doesn't work for me. It says Serial Number: and then nothing. cfgadm -alv seems to work, and SMARTD works well enough, though I have to massage it to read the disks that aren't on my SAS controller.

It only shows the first 6 or so digits printed on the label so it's enough to ID the old drives.

I'm not as sure about the new drives because, for better or worse, the serial numbers span about a thousand #s so the first few digits are the same.

don't gently caress me again WD... :tinfoil:

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

For me it just shows nothing, and even if it was a few digits, that wouldn't help me as the first 15 or so digits of my Samsung drives are the same.
code:
c7t0d0           Soft Errors: 0 Hard Errors: 0 Transport Errors: 0 
Vendor: ATA      Product: SAMSUNG HD154UI  Revision: 1118 Serial No:  
Size: 1500.30GB <1500301910016 bytes>
Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 0 
Illegal Request: 26 Predictive Failure Analysis: 0 

Weird, maybe it's not supported by your controller?
code:
Model: WDC WD3200AAKS- Revision:  Serial No:      WD-WCAPZ25 Size: 320.07GB <320070352896 bytes>
Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 0 
Illegal Request: 0 
also, it may matter:
code:
# uname -a
SunOS repository 5.11 snv_98 i86pc i386 i86pc Solaris
EDIT: gently caress you WD. Drive 3/4 resilvered fine, but is starting to show read errors while resilvering #4. #4 is showing write errors now! It better finish resilvering fine!!!!!

code:
        NAME           STATE     READ WRITE CKSUM
        tank           DEGRADED     0     0     0
          raidz1       DEGRADED     0     0     0
            c1d0       ONLINE       2     0     0
            replacing  DEGRADED     0     4     0
              c5d1     OFFLINE      0     0     0
              c1d1     ONLINE       0     6     0
            c2d0       ONLINE       0     0     0
            c2d1       ONLINE       0     0     0
EDIT2: :sigh:
code:
  pool: tank
 state: DEGRADED
status: One or more devices has experienced an unrecoverable error.  An
        attempt was made to correct the error.  Applications are unaffected.
action: Determine if the device needs to be replaced, and clear the errors
        using 'zpool clear' or replace the device with 'zpool replace'.
   see: [url]http://www.sun.com/msg/ZFS-8000-9P[/url]
 scrub: scrub in progress for 0h6m, 1.99% done, 5h41m to go
config:

        NAME           STATE     READ WRITE CKSUM
        tank           DEGRADED     0     0     0
          raidz1       DEGRADED     0     0     0
            c1d0       ONLINE       6     0     0
            replacing  UNAVAIL      0    12     0  insufficient replicas
              c5d1     OFFLINE      0     0     0
              c1d1     FAULTED      0    18     0  too many errors
            c2d0       ONLINE       0     0     0
            c2d1       ONLINE       0     0     0

Delta-Wye fucked around with this message at 06:56 on Jul 25, 2012

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

There's no data loss. He's trying to replace drives to get a bigger pool while both drives are online, but it looks like the old drive he's replacing crapped out before it could copy over to the new one. He can still rebuild the new one from the rest of the drives since it's RAIDZ.

And if there's data loss, ZFS will tell you what files are affected. I had 3 unrecoverable errors in my array, and I know which two files it was (and have dealt with those).

I would be totally aok with data loss across like 4/5s of the drive, and the other 5th is packed up so whatev. And, being the badass motherfucker zfs is I slapped the old drive in, scrubed, and canceled the replacement via a detach and:

code:
  pool: tank
 state: ONLINE
 scrub: scrub completed after 9h3m with 0 errors on Wed Jul 25 07:11:34 2012
config:

        NAME        STATE     READ WRITE CKSUM
        tank        ONLINE       0     0     0
          raidz1    ONLINE       0     0     0
            c1d0    ONLINE       0     0     0
            c5d1    ONLINE       0     0     0
            c2d0    ONLINE       0     0     0
            c2d1    ONLINE       0     0     0

errors: No known data errors
It's all good! :toot:

...Or I have silent data errors... :smith:

Currently the drive I suspect is bad is in my desktop getting reamed with a dd and then a smart diagnostic test. Is there a better way to stress test a new drive? Google gave me a bunch of sketchy ideas that aren't much better than pushing the disk with a dd operation and observing for errors.

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

Ain't no such thing, ZFS keeps a checksum of all the files, that's how it knows there are no data errors, the scrub checks every block to make sure it's still good.

I would think so too, but I'm not sure how it's physically possible. I had three out of four disks in, rebuilding the fourth, and it started having read errors on one of the three remaining. As far as I understand it there shouldn't have been enough information available to rebuild without error? I don't know; could be there were recoverable errors. Or the write errors were the root of the problem? Now that I think the fire is put out I'm going to have to spend some time tonight debugging which drive had the problem.

Delta-Wye
Sep 29, 2005

evil_bunnY posted:

Read errors just means the checksum and the data weren't consistent. It can try again.
In any case, this poo poo is why you run R6/RZ2.

Is there any way to 'promote' a RZ1 to an RZ2 with the addition of additional disks? I am stuck in a 4-bay NAS case at the moment but interested in replacing that already anyways. EDIT: Survey says.. no! :(

FYI I guess you can plug in the 'new drive' and do the replacement while the old drive is still in place so poo poo doesn't go tits up while you're doing so but I don't have space and I am worried the power supply won't support it (it's actually on my list of things that might have caused this last problem).

Delta-Wye fucked around with this message at 19:50 on Jul 25, 2012

Delta-Wye
Sep 29, 2005
I'm trying to debug what's going on with this last drive. Currently, everything in the NAS looks fine except I have one last 320GB drive I'd love to swap out. I've tried a couple different things with the 'bad' drive, including swapping it for one of the drives that seemed to work. The swap would /not/ go through and kept giving me arcane errors like:
code:
# zpool replace tank c1d0
cannot replace c1d0 with c1d0: new device must be a single disk
:iiam:
smart tools on solaris seem to be junk so I swapped the drive over to my linux box and the only issue that seems to crop up is:
199 UDMA_CRC_Error_Count 0x0032 200 200 000 Old_age Always - 5
The first attempt made it go to 2 or 3, second attempt to 5. Googling indicates that is usually a cabling issue of some sort and not critical. I would like to get this drive to freak out so that I can have some excuse to RMA it pronto but on my Linux box where I feel comfortable working, I get no errors at all.

:sigh:

Any ideas?

EDIT: Also, dmesg on the NAS shows a ton of crap like
code:
Jul 25 20:49:28 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:28 repository      timeout: early timeout, target=0 lun=0
Jul 25 20:49:28 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:28 repository      timeout: abort request, target=1 lun=0
Jul 25 20:49:28 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:28 repository      timeout: abort device, target=1 lun=0
Jul 25 20:49:28 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:28 repository      timeout: reset target, target=1 lun=0
Jul 25 20:49:28 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:28 repository      timeout: reset bus, target=1 lun=0
Jul 25 20:49:29 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:29 repository      timeout: early timeout, target=1 lun=0
Jul 25 20:49:29 repository scsi: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0 (ata2):
Jul 25 20:49:29 repository      timeout: early timeout, target=0 lun=0
Jul 25 20:49:29 repository gda: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0/cmdk@0,0 (Disk1):
Jul 25 20:49:29 repository      Error for command 'read sector' Error Level: Informational
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Sense Key: aborted command
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Vendor 'Gen-ATA ' error code: 0x3
Jul 25 20:49:29 repository gda: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0/cmdk@1,0 (Disk2):
Jul 25 20:49:29 repository      Error for command 'write sector'        Error Level: Informational
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Sense Key: aborted command
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Vendor 'Gen-ATA ' error code: 0x3
Jul 25 20:49:29 repository gda: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0/cmdk@1,0 (Disk2):
Jul 25 20:49:29 repository      Error for command 'write sector'        Error Level: Informational
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Sense Key: aborted command
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Vendor 'Gen-ATA ' error code: 0x3
Jul 25 20:49:29 repository gda: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0/cmdk@0,0 (Disk1):
Jul 25 20:49:29 repository      Error for command 'read sector' Error Level: Informational
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Sense Key: aborted command
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Vendor 'Gen-ATA ' error code: 0x3
Jul 25 20:49:29 repository gda: [ID 107833 kern.warning] WARNING: /pci@0,0/pci-ide@9/ide@0/cmdk@1,0 (Disk2):
Jul 25 20:49:29 repository      Error for command 'write sector'        Error Level: Retryable
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Sense Key: ICRC error during UDMA
Jul 25 20:49:29 repository gda: [ID 107833 kern.notice]         Vendor 'Gen-ATA ' error code: 0x16
Is this green drive related TLER poo poo?

Delta-Wye fucked around with this message at 08:39 on Jul 26, 2012

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

What kind of drive is it? Is it a Seagate 1.5 TB by any chance?

the four new ones are WD Green 2TB EARX models. The ones generating read and write errors are on the same back plane, but I didn't have any trouble with that with the other two on the other controller.

Delta-Wye
Sep 29, 2005
I've been sort of dinking around with my new drives in my spare time and still can't get it working. It's frustrating because neither drive separately show errors, but together they are very very unhappy. At this point, I actually suspect it may be a power issue; the new drives draw a little bit more current than the old ones according to the labels and it could be that all four of the new drives might be a bit too much.

I need to replace the case anyways, so I've begun hunting around for a replacement for a small NAS. As expected, all of the nice cases are pretty expensive so I figured if I'm going to go all out and rebuilding from scratch I might as well get two more drives and switch to raidz2. So I'm looking at a couple miniATX cases that support that many drives (but none with sweet drive trays like my old Chenbro) along with a miniATX board that can handle that many drives. This has turned out to be pretty difficult!

Cases seem limited to Lian Li PC-Q08B, PC-Q25B and the Fractal Design Array R2.

The motherboard is even more of a PITA. Very few have 7ish SATA connection (right now I have one with 4 for the storage drives and a PATA connection for the system drive) and I don't know if the cases will have room for a large card like the M1015.

Are there any common builds for such a computer? It doesn't seem as popular as I would have guessed.

Delta-Wye
Sep 29, 2005

Rukus posted:

I "upgraded" from a PC-Q08B to a PC-Q25B and I really like it. I'm running my drives in a JBOD pool, so the card I use is a SuperMicro SAS to SATA: http://www.supermicro.com/products/accessories/addon/aoc-sas2lp-mv8.cfm. It works great, and supports 3TB drives (and probably 4TB as well).

Like b0lt mentioned, definitely look for a proper-fitting PSU. Modular is preferred, but the recommended manufacturers' (Seasonic/Corsair/XFX) modular PSUs are too long. I ended up using a Corsair 400W (the older builder series, when they were still Seasonic rebadges) and just placed the extra cables under the PSU.

Another option is Silverstone's Strider PSU: http://www.newegg.com/Product/Product.aspx?Item=N82E16817256065. You can also pick up shorter cables for it, too: http://www.newegg.com/Product/Product.aspx?Item=N82E16812162010. Though I've heard that it uses cheaper capacitors and the fan is a bit more audible. It's a tradeoff for being fully modular for that size of PSU.

Thanks for the info guys. Do you use a freebsd varient with that SATA card? Driver support is another issue of concern :(

Also, my fellow packrats... Newegg has WD20EARX 2TB drives for $100 again: http://www.newegg.com/Product/Product.aspx?Item=N82E16822136891

Delta-Wye
Sep 29, 2005
No solaris doesn't mean no *bsd, right?

Just checking...

Delta-Wye
Sep 29, 2005

evil_bunnY posted:

Different kernel, BSD ran on a wider range of hardware last time I checked.

I do know the difference between the two, and I am aware of BSDs pretty decent hardware support; I am asking about this particular hardware. If it works I may be ordering it next day or two. If this particular item doesn't (and usually it's the cheaper lowend equipment that has the spottiest support) then I will keep digging for a sata solution.

Delta-Wye
Sep 29, 2005

sellouts posted:

How are you connecting to and transferring the files to the NAS?

Edit: Is anyone buying a special enclosure for their set up or just using a spare case that they have around?

I don't have any spare cases and want 5+ drive bays, hot swap would be nice, and a level of RAID that provides some level of redundancy. I'm sharing media and doing backups from a few computers, nothing crazy.

Is there a solid case for this? Or would I be better off by just going with one of the off the shelf solutions?

I rolled my own previously using a Chenbro NAS case and liked it but it got all smashed up so I'm going to get a LianLi PC-Q25B and roll my own next. There are a ton of posts on the previous page with my questions and people chiming in.

To be honest, I think it's probably more effort than it's worth to make your own vs. using an off-the-shelf unit but I enjoy it so it's personally worth it to me. If I wanted a simple effective turnkey solution slapping one together myself would be a bit silly.

Delta-Wye
Sep 29, 2005
Maybe it's just me, but too many Calvin and Hobbes avatars leaves this conversation very difficult to follow :(

Delta-Wye
Sep 29, 2005
I am ever-so-slowly moving to getting my NAS upgrade finished. At this point, it is clear I will be rebuilding the machine from scratch. New case and new power supply so far, along with 5x 2TB drives. My old motherboard only has four SATA ports, so I either need a new motherboard or one of the SAS cards discussed a few pages ago.

I looked into the cards, and they are pretty much all pci-express. My current motherboard has a PCI slot, so I am stuck replacing the motherboard too. I figure i might as well try and get something with more SATA ports and save myself the trouble and cost of the additional SAS card. My current system disk is a PATA disk, so that will need to replaced too!

This whole project has gotten much more expensive than just "replace case and harddisks" but I think it will be for the best. I would like to turn it into a hybrid HTPC/NAS and found a mini-itx motherboard with 5x SATA ports and HDMI out. I want to slap FreeBSD on it to use ZFS and probably some sort of media management software (I'll get that figured out down the road). The board in question is the Asus E45M1-I Deluxe which has every feature I need plus some, but software support may be spotty.

The video card, at the very least, will probably only work in VESA mode. The SATA ports are 6Gb/s, so they should work nicely. I like the passive cooling; the current motherboard's fans are so. damned. annoying. The case has some larger dimension fans that I'm hoping will push enough air to keep the board reasonably cool at a much lower volume. It's also low power, which is good in the long term. In theory, it includes the processor, enough SATA ports, usable GigE network port, video card, etc, so this and 8GB of RAM may be my last purchase for this project. One last expensive purchase, but it will be over with.

A few questions:

Does anyone have any personal experience with this board? I imagine the processor will be enough for ZFS, but doing ZFS/background tasks like sabnzbd/playing media all at the same time might be pushing it. Any thoughts?

There is no PATA connector and I'm planning on using all 5 SATA ports for the storage disks, meaning I need a solution for my boot medium. Am I stupid for considering something like a thumb drive and the USB 3.0 port for my boot partition? FreeBSD seems to support it, and I figure I can probably move most of the active stuff (logging, etc) to a partition on the storage drives if I'm worried about write cycles on a thumb drive. This used to be a concern, but I'm not sure how big of a deal it is anymore. Hell, with the cost of thumb drives I can easily keep a snapshot of the boot image around and just replace it every once in a while.

Delta-Wye
Sep 29, 2005

movax posted:

For my most recent upgrade to my box, I bought a used X58 motherboard, used i7-920 and 24GB of RAM all for a little less than $400, if I recall correctly. The RAM was the only part I got new. I'd say taking advantage of frequent upgraders and buying their used gear is the way to go.

Not a bad idea, but my needs are so odd that it's hard. For instance, I need a shitload of SATA ports on a mini-itx board. It's not easy being green.

Longinus00 posted:

PCI has a max bandwidth (assuming 32bit @ 33Mhz) of around 133MB/s so don't stick more than 1 disks on their if you want max throughput. Obviously if your bottleneck is your processor then you can probably get away with more.
Er, poorly stated. I figured the SAS card would be the solution so I looked into it, then realized it wasn't compatible with my motherboard, not that I went hunting for a PCI SAS card in the first place. Mostly a problem because my dumb self didn't memorize the features of my NAS box, it's been really good to me in regards to just working since i put it together.

Longinus00 posted:

For a HTPC you *really* want hardware assisted decoding. Have you researched how well this is going to work in FreeBSD?
Probably not at all! The HTPC aspect is very much a "well, that would be nice," mostly because even if I get everything else working, FreeBSD won't support Silverlight so I can't run Netflix. I don't care, but the SO would be disappointed and it would be mostly useless for her.

Longinus00 posted:

I think some cheap thumbdrives might not have any wear leveling at all so you'll want to use a filesystem made for flash in those cases (e.g. linux's logfs).

I think in general you might need to do more research into what you're planning to do. Particularly regarding how you're actually going to go about doing the HTPC side of things.
I think I've got the NAS end of it pretty situated. I'm really just stuck on the usb3 boot device; good idea, bad idea, or merely doable?

Delta-Wye
Sep 29, 2005

FISHMANPET posted:

Trying to build a NAS plus HTPC is a fools errand. A $100 Roku box can deal with basically every online streaming service ever. A $200 HTPC (E-350/E-450 based) will play all the content off your NAS, and be as quiet as a whisper.

That may end up being what I do (if I do anything on the HTPC front) so it's not a big deal if I can't coerce the NAS into doing it the way I wish; a barebones HTPC would be pretty cheap, especially if I go all out with removing unnecessary features like harddisks, etc. It's just the NAS is already on all the time, and at the moment will probably be parked next to the TV anyways so double-duty is ideal. However, it is pretty clear that ZFS and netflix are not going to run on the same machine without something relatively insane like a VM solution.

Delta-Wye
Sep 29, 2005

Longinus00, a bit ago posted:

Were you using the RAID functionality of your Intel card or doing softraid? If you were using the on card RAID then you will probably need another RT3WB080 (maybe if you're lucky another card with the same chipset might work). If you were running any form of softraid then it doesn't matter how you connect the disks so long as the cards you connect them with don't try to "initialize" them.

This is why I would recommend against a hardware raid solution of any level. Are you talking about a hardware raid5? Or a mdadm raid5?

At some level there isn't much difference between a software raid5 and a software zfs array based on the problems you're having. In theory the poor performance might be mitigated by offloading checksums to a hardware raid card, but your hardware should be sufficient for the bulk of the calculations which makes me suspect something else is going on.

The lockups sound like flaky hardware and/or flaky drivers, both of which can be fixed by swapping to better supported hardware and upgrading to a new operating system without any changes to your storage setup.

Delta-Wye
Sep 29, 2005

fletcher posted:

Why oh god? If I go that route instead, the NAS won't be able to get to out to the internet right?

On windows, i would do Internet Connection Sharing. It will set your desktop up as a DHCP server, assign a private address to the wired connection. NAS4FREE will grab an address and the desktop will route everything properly.

Adbot
ADBOT LOVES YOU

Delta-Wye
Sep 29, 2005

astr0man posted:

So I'm dumb and just discovered that 512b vs 4k sector sizes is a thing. I have a microserver with 4 512b sector drives and a zpool that's configured with ashift=9. I have 4 new advanced format drives, and just zpool replace'ing 1 at a time won't work because it complains about the non-matching sector sizes.

From a bit of googling it seems like I have to create a new pool with ashift=12 for the new drives, but since I only have 1 microserver I have no idea how to get all my data copied into the new pool.

I guess I could plug the 4 new drives into my desktop and set up the new pool there, but if there's an easier way I'd love to hear it.

Similar question, but from a different angle. I'm setting up a zpool from scratch, and I'm not sure which ashift value is 'right for me'. I can set each one up one at a time and test it but what should I be testing?

EDIT: I'm running WD20EARX (yes, green drives :sigh:) x5 in Raidz2 configuration if that helps narrow the answers down. It will probably be containing >>4k datafiles, not tons of tiny text files.

Delta-Wye fucked around with this message at 07:50 on Sep 29, 2012

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply