Hi all! I feel i should start with some back story for this project. (sorry in advance. After typing this i realized i should maybe be a novelist -.- I sincerely tried to google and forum search for as many answers as I could. I read the manuals, I learned a lot in the last few days, but still have some newb problems and it is ALL new territory to me.)
I'm sorta new to this whole FreeNAS thing. I've been using it for about 8 months now but have only had time to learn about the things that are failing on it or that I want to expand. :P
I read about FreeNAS while looking at a solution to alleviate my router-attached dlna problems (no surprise; they suck). The guide i read was on FreeNAS 7 and 9.x was already out, so I had some serious learning curves. I read about miniDLNA and wanted to try that (I have since moved to PLEX and it's the greatest thing I've ever ever). I originally tried to salvage an old computer from my garage to use (it was one of THOSE blogs), but after 5 hours of countless problems including not even being able to INSTALL without kernel panics, I said "Screw it!!" (with many superlatives) and jumped on New Egg and built something that was cheap (I was/am kinda broke).
So. Originally it was a budget build (pretty much completely with un-suggested hardware[i know this now]) that ran me about $300. I had a lot of experience with/building desktops and none with servers (that has changed a lot in the last 6 months). This was the original build:
Haswell Celeron Processor(then i read to not use haswell)
MSI H81M-E35 Mobo (not ECC and has a Realtek NIC)
Mushkin redline 1600MHz memory (not ECC)
A 3TB seagate backup plus drive i had lying around (USB = FAIL)
A very small microATX case with integrated PSU (purchased because small and usb3.0 for my fail drive)
I simultaneously started piecing together a gaming desktop and then got a better job that let me expedite the desktop build (and buy better hardware). So, i had an extra mid tower lying around. Also around this time my FreeNAS box was starting to give me problems.
Originally i spent hours learning how to bumble around Shell and the file system. I installed miniDLNA and configured it. I set up a custom jail and installed an Apache web server with php and mysql, etc (THAT was an experience and entirely shell). I then read up on Plex, tried it out, and immediately threw MiniDLNA out the window. However, I started getting problems with Plex one day where I couldnt access my movie, then the server. I had it headless at the time, so i hard shut it down and went to work.
I came home later and had more and more problems accessing the server. It would boot and the web GUI would never be accessible. Finally i hooked it up to a monitor and i was getting tons of disk mount errors on the printout. (yay usb). I figured maybe it was the crap USB interface, so I decided to rip the drive out of the external enclosure and put it internal. There was no room in the crappy uATX box, so I moved the whole build to the mid tower I had lying around (looks better anyway and runs cooler). While moving it, I discovered that the PSU cables (in their tight configuration) nearly ripped the USB3.0 header off the mobo (no wonder I had issues...). I put everything in the box and booted and more problems (UGH). First issue was drive recognition. The SATA III cables I bought from Fry's somehow didn't support AHCI mode. Fine. I used some old SATA II cables (the HDDs cant i/o fast enough to matter) and AHCI worked again. Still drive issues, can't mount, zpool failure blah.
Then i thought "maybe i'll unmount the zpool and try to import it again". Well, that didn't go well... I got FreeNAS to read the drive again, but it wouldnt import the volume, so I had to create a new one (thankfully I had backed up 2 weeks sooner before all this happened). I realized that this point that I needed to have some parity capabilities so I decided to go with a 3-drive RaidZ1. I ordered a new drive, backed up my original backup drive to another external, and used drives 1/2 and the new drive to build the RaidZ1 array (All barracuda drives). So, I tried to write my backup to the disk (2.5 TB of the 6TB capacity) and it failed halfway through. Zpool was degraded, and I was getting all kinds of mounting problems printing out on the monitor for the original HDD (the old USB one). So, I scrapped that drive and decided to stop buying crap hardware, so I got a 3TB Seagate NAS drive to replace it.
Got the new NAS drive, deleted the zpool (i figured that was easiest and i still had the backup), made a new RaidZ1 volume with the NAS drive and 2 Barracuda drives, and wrote my data. I reconfigured PLEX and my CIFS share and was back on my way.
WednesdayI jumped on the WebGUI and had a yellow light (not the end of the world). It said I had a drive error on one of the Barracuda drives, so I restarted the machine and started a resilver of the bad drive. A few minutes later I got an Unrecoverable Error message for the drive, so I took the server offline, shut it down, and ordered another NAS drive. (thankfully I live in CA, so Super EggSaver shipping is overnight for me). Yesterday I pulled out what I thought was the bad drive, and popped in the new NAS drive. When i booted the box it showed a cks error on a drive still and a red light Unknown Size error on the volume, so i shut it back down, put the "bad" drive back in and booted. Then i decided to do more research on scrubbing (didn't know what it was), S.M.A.R.T testing (had never used them) and decided to run some tests.
I scrubbed the volume (fixed 800+GB of errors) and ran short SMART tests on the drives (they all passed). I then was able to use zpool clear through shell (after researching how/what it does) and resilvered the drive (successfully). Today I rescrubbed the drive with no errors and looked up how to use smartctl to initiate a manual long test (it still has 50% to go).
Now that I have a novella, to the questions:
Is my zpool in trouble and what is making 2 drives fail in such a short period of time? I will know after the long test more about the problems with the drive. As long as it isn't falling out of the pool consistently or throwing SMART failures should I be fine with the occasional scrub to fix issues (I scrub weekly now, should I do daily?)
Is my mobo/RAM to blame for this? am I getting bitten by my lack of ECC? I sincerely think the original drive went bad. I had it on USB and set to spin down to save power (which i now know I shouldn't do). But a second drive (and a new one at that?).
Additionally, I am faced with some other questions that are spawned by my 48-hour, deeper dive into FreeNAS/FreeBSD:
1. When I add a new ZFS volume is that a new zPool? I decided to use that extra NAS drive i ordered as a backup drive for my desktop (File History and weekly scheduled backups of C:\). I read Most of Cyberjock's presentation on Zpools and vdevs and want to make sure I didn't screw my zpool? or do I have 2 now?
2. I thought about getting a RAID card instead of screwing with my H81 chipset for drives. I was looking at an LSI 9240-4i since I know LSI makes good raid cards, but now after reading the hardware list I don't see LSI on there and I see 800 million Adaptec controllers. Should I go with them? Should I use a RAID controller? can I do ZFS with a raid card or is hardware raid a better option? I'm not sure how I would go about configuring that with my current software? raid option. (also I assume I would have to back up my data to external drives before i install a RAID card).
3. I'm having a screwy issue with CIFS where I can see the server in network locations, but half the time I get a connection error from explorer.exe when opening the server, the other half I get the connection problem when trying to open the shared drive. (windows cannot access //YOURSERVER. Check the spelling of the name. Otherwise network blah might blah blah). Either way this is usually fixed by logging into the WebGUI, or in extreme cases from using FileZilla to SFTP into the server and opening the /mnt/ folder. (i just use network auth to log into the CIFS share either with root or a single user). I thought this was a problem originally because my USB disk was spinning down, but it has continued across new zpools, new drives, and a switch to constant spin-up. Maybe it's a cstates issue? I'm not sure if I have them turned on. Maybe it's my realtek NIC falling asleep and it wakes on port 22 but not the CIFS port? [maybe i should start a new thread for this one]
Oh man. That's enough for now. I commend the brave soul who actually read this. I know my build is janky. I use it as a file server/DLNA server only, so I'm not going to put $3k into server hardware (at least unless it keeps pissing me off and i get a bug up my ass and a dangerous desire to use that credit card)
Thanks for any help that anyone can provide
I'm sorta new to this whole FreeNAS thing. I've been using it for about 8 months now but have only had time to learn about the things that are failing on it or that I want to expand. :P
I read about FreeNAS while looking at a solution to alleviate my router-attached dlna problems (no surprise; they suck). The guide i read was on FreeNAS 7 and 9.x was already out, so I had some serious learning curves. I read about miniDLNA and wanted to try that (I have since moved to PLEX and it's the greatest thing I've ever ever). I originally tried to salvage an old computer from my garage to use (it was one of THOSE blogs), but after 5 hours of countless problems including not even being able to INSTALL without kernel panics, I said "Screw it!!" (with many superlatives) and jumped on New Egg and built something that was cheap (I was/am kinda broke).
So. Originally it was a budget build (pretty much completely with un-suggested hardware[i know this now]) that ran me about $300. I had a lot of experience with/building desktops and none with servers (that has changed a lot in the last 6 months). This was the original build:
Haswell Celeron Processor(then i read to not use haswell)
MSI H81M-E35 Mobo (not ECC and has a Realtek NIC)
Mushkin redline 1600MHz memory (not ECC)
A 3TB seagate backup plus drive i had lying around (USB = FAIL)
A very small microATX case with integrated PSU (purchased because small and usb3.0 for my fail drive)
I simultaneously started piecing together a gaming desktop and then got a better job that let me expedite the desktop build (and buy better hardware). So, i had an extra mid tower lying around. Also around this time my FreeNAS box was starting to give me problems.
Originally i spent hours learning how to bumble around Shell and the file system. I installed miniDLNA and configured it. I set up a custom jail and installed an Apache web server with php and mysql, etc (THAT was an experience and entirely shell). I then read up on Plex, tried it out, and immediately threw MiniDLNA out the window. However, I started getting problems with Plex one day where I couldnt access my movie, then the server. I had it headless at the time, so i hard shut it down and went to work.
I came home later and had more and more problems accessing the server. It would boot and the web GUI would never be accessible. Finally i hooked it up to a monitor and i was getting tons of disk mount errors on the printout. (yay usb). I figured maybe it was the crap USB interface, so I decided to rip the drive out of the external enclosure and put it internal. There was no room in the crappy uATX box, so I moved the whole build to the mid tower I had lying around (looks better anyway and runs cooler). While moving it, I discovered that the PSU cables (in their tight configuration) nearly ripped the USB3.0 header off the mobo (no wonder I had issues...). I put everything in the box and booted and more problems (UGH). First issue was drive recognition. The SATA III cables I bought from Fry's somehow didn't support AHCI mode. Fine. I used some old SATA II cables (the HDDs cant i/o fast enough to matter) and AHCI worked again. Still drive issues, can't mount, zpool failure blah.
Then i thought "maybe i'll unmount the zpool and try to import it again". Well, that didn't go well... I got FreeNAS to read the drive again, but it wouldnt import the volume, so I had to create a new one (thankfully I had backed up 2 weeks sooner before all this happened). I realized that this point that I needed to have some parity capabilities so I decided to go with a 3-drive RaidZ1. I ordered a new drive, backed up my original backup drive to another external, and used drives 1/2 and the new drive to build the RaidZ1 array (All barracuda drives). So, I tried to write my backup to the disk (2.5 TB of the 6TB capacity) and it failed halfway through. Zpool was degraded, and I was getting all kinds of mounting problems printing out on the monitor for the original HDD (the old USB one). So, I scrapped that drive and decided to stop buying crap hardware, so I got a 3TB Seagate NAS drive to replace it.
Got the new NAS drive, deleted the zpool (i figured that was easiest and i still had the backup), made a new RaidZ1 volume with the NAS drive and 2 Barracuda drives, and wrote my data. I reconfigured PLEX and my CIFS share and was back on my way.
WednesdayI jumped on the WebGUI and had a yellow light (not the end of the world). It said I had a drive error on one of the Barracuda drives, so I restarted the machine and started a resilver of the bad drive. A few minutes later I got an Unrecoverable Error message for the drive, so I took the server offline, shut it down, and ordered another NAS drive. (thankfully I live in CA, so Super EggSaver shipping is overnight for me). Yesterday I pulled out what I thought was the bad drive, and popped in the new NAS drive. When i booted the box it showed a cks error on a drive still and a red light Unknown Size error on the volume, so i shut it back down, put the "bad" drive back in and booted. Then i decided to do more research on scrubbing (didn't know what it was), S.M.A.R.T testing (had never used them) and decided to run some tests.
I scrubbed the volume (fixed 800+GB of errors) and ran short SMART tests on the drives (they all passed). I then was able to use zpool clear through shell (after researching how/what it does) and resilvered the drive (successfully). Today I rescrubbed the drive with no errors and looked up how to use smartctl to initiate a manual long test (it still has 50% to go).
Now that I have a novella, to the questions:
Is my zpool in trouble and what is making 2 drives fail in such a short period of time? I will know after the long test more about the problems with the drive. As long as it isn't falling out of the pool consistently or throwing SMART failures should I be fine with the occasional scrub to fix issues (I scrub weekly now, should I do daily?)
Is my mobo/RAM to blame for this? am I getting bitten by my lack of ECC? I sincerely think the original drive went bad. I had it on USB and set to spin down to save power (which i now know I shouldn't do). But a second drive (and a new one at that?).
Additionally, I am faced with some other questions that are spawned by my 48-hour, deeper dive into FreeNAS/FreeBSD:
1. When I add a new ZFS volume is that a new zPool? I decided to use that extra NAS drive i ordered as a backup drive for my desktop (File History and weekly scheduled backups of C:\). I read Most of Cyberjock's presentation on Zpools and vdevs and want to make sure I didn't screw my zpool? or do I have 2 now?
2. I thought about getting a RAID card instead of screwing with my H81 chipset for drives. I was looking at an LSI 9240-4i since I know LSI makes good raid cards, but now after reading the hardware list I don't see LSI on there and I see 800 million Adaptec controllers. Should I go with them? Should I use a RAID controller? can I do ZFS with a raid card or is hardware raid a better option? I'm not sure how I would go about configuring that with my current software? raid option. (also I assume I would have to back up my data to external drives before i install a RAID card).
3. I'm having a screwy issue with CIFS where I can see the server in network locations, but half the time I get a connection error from explorer.exe when opening the server, the other half I get the connection problem when trying to open the shared drive. (windows cannot access //YOURSERVER. Check the spelling of the name. Otherwise network blah might blah blah). Either way this is usually fixed by logging into the WebGUI, or in extreme cases from using FileZilla to SFTP into the server and opening the /mnt/ folder. (i just use network auth to log into the CIFS share either with root or a single user). I thought this was a problem originally because my USB disk was spinning down, but it has continued across new zpools, new drives, and a switch to constant spin-up. Maybe it's a cstates issue? I'm not sure if I have them turned on. Maybe it's my realtek NIC falling asleep and it wakes on port 22 but not the CIFS port? [maybe i should start a new thread for this one]
Oh man. That's enough for now. I commend the brave soul who actually read this. I know my build is janky. I use it as a file server/DLNA server only, so I'm not going to put $3k into server hardware (at least unless it keeps pissing me off and i get a bug up my ass and a dangerous desire to use that credit card)
Thanks for any help that anyone can provide
