RaidCore/BroadCom BC4852 8 port compatibility?

[Joe L.]

I downloaded the free version of unraid to a usb drive and booted my current machine.  From the web interface, I was able to see all the drives attached to the BC4852 RaidCore card.  I am not ready to initialize my unraid array yet, so I didn't do anything other than "look".  I posted a screen shot and syslog in another thread I have going in case it gives any clues on whether the BC4852 is going to work or not.  Here's the post with the attachments.

 

http://lime-technology.com/forum/index.php?topic=5185.msg48082#msg48082

 

thanks,

Murray

Based on the screen shot, it looks like it will work.

 

The syslog on the other post indicates the kernel detected a BIOS bug, where some memory is not mapped,  but worked around it.  You might check and see if a BIOS update is available.  (don't discount the board if one is not, as the same buggy BIOS is running your current array just fine.)  Other than that, I did not see anything that looked odd.

 

Joe L.

[WeeboTech]

It has promise !!!! (I stand corrected). Nice, if this works well, I bet it could bring new life into used ones from eBay.

 

Jan 26 11:08:08 Tower kernel: sata_svw 0000:06:02.0: version 2.3

Jan 26 11:08:08 Tower kernel: sata_svw 0000:06:02.0: PCI INT A -> GSI 18 (level, low) -> IRQ 18

Jan 26 11:08:08 Tower kernel: scsi0 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi1 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi2 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi3 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi4 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi5 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi6 : sata_svw

Jan 26 11:08:08 Tower kernel: scsi7 : sata_svw

 

 

[WeeboTech]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

[Joe L.]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

Most scripts and examples use this syntax (with added "-d ata" option):

smartctl -a -d ata /dev/sda

[MurrayW]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

Most scripts and examples use this syntax (with added "-d ata" option):

smartctl -a -d ata /dev/sda

Do I run this from the actual server through the command line or from the web interface?  I am guessing it is from the server.  If that is correct, how do I capture the log file? 

 

Thanks for your help.

 

Murray

[Joe L.]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

Most scripts and examples use this syntax (with added "-d ata" option):

smartctl -a -d ata /dev/sda

Do I run this from the actual server through the command line or from the web interface?  I am guessing it is from the server.  If that is correct, how do I capture the log file? 

 

Thanks for your help.

 

Murray

Several methods possible...

 

You can do this from your PC.  Log onto your server from the PC by using "telnet" (Login as "root", press enter when prompted for a password.  Run the smartctl command, Then cut/paste the results from the telnet window.)

 

Or you can do it from the system console capturing the output to a file in /var/log

smartctl -a -d ata /dev/sda 2>&1 | tee /var/log/smart_sda.txt

Then, you can access the file in /var/log from your browser by browsing to:

http://tower/log/smart_sda.txt

 

Joe L.

[MurrayW]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

Most scripts and examples use this syntax (with added "-d ata" option):

smartctl -a -d ata /dev/sda

Do I run this from the actual server through the command line or from the web interface?  I am guessing it is from the server.  If that is correct, how do I capture the log file? 

 

Thanks for your help.

 

Murray

Several methods possible...

 

You can do this from your PC.   Log onto your server from the PC by using "telnet" (Login as "root", press enter when prompted for a password.  Run the smartctl command, Then cut/paste the results from the telnet window.)

 

Or you can do it from the system console capturing the output to a file in /var/log

smartctl -a -d ata /dev/sda 2>&1 | tee /var/log/smart_sda.txt

Then, you can access the file in /var/log from your browser by browsing to:

http://tower/log/smart_sda.txt

 

Joe L.

Thanks...I'll do it tonight when I get home and post the results.

[MurrayW]

Another good test would be to see how the kernel handles grabbing SMART information.

You do not have to assign a drive to the unRAID environment.

Boot up, login as root.

 

run

smartctl -a /dev/sda

 

see if it grabs a log successfully without any additional switches. I know the AMC/3ware cards require a special switch.

Most scripts and examples use this syntax (with added "-d ata" option):

smartctl -a -d ata /dev/sda

Do I run this from the actual server through the command line or from the web interface?  I am guessing it is from the server.  If that is correct, how do I capture the log file? 

 

Thanks for your help.

 

Murray

Several methods possible...

 

You can do this from your PC.   Log onto your server from the PC by using "telnet" (Login as "root", press enter when prompted for a password.  Run the smartctl command, Then cut/paste the results from the telnet window.)

 

Or you can do it from the system console capturing the output to a file in /var/log

smartctl -a -d ata /dev/sda 2>&1 | tee /var/log/smart_sda.txt

Then, you can access the file in /var/log from your browser by browsing to:

http://tower/log/smart_sda.txt

 

Joe L.

Thanks...I'll do it tonight when I get home and post the results.

I telnetted in and kept getting errors when using the -d ata option.  The attached log was run using smartctl -a /dev/sda

 

Thanks again for all your help.

 

Murray

smart.txt

[MurrayW]

Any of you experts have a chance to look at my log from my previous post to see if it tells you anything?

 

thanks,

Murray

[WeeboTech]

Any of you experts have a chance to look at my log from my previous post to see if it tells you anything?

 

thanks,

Murray

 

It tells us that the drive is directly accessible to acquire SMART data.

This means that emhttp might be able to access the drive directly for hdd temperature and status (a big plus).

Can you post what the error was when you used -d ata (that should have worked).

 

So far this is looking very promising.

 

[MurrayW]

Any of you experts have a chance to look at my log from my previous post to see if it tells you anything?

 

thanks,

Murray

 

It tells us that the drive is directly accessible to acquire SMART data.

This means that emhttp might be able to access the drive directly for hdd temperature and status (a big plus).

Can you post what the error was when you used -d ata (that should have worked).

 

So far this is looking very promising.

 

I'll run the -d ata again this weekend and post the exact error.  It was something like it was an unsupported command and referred me to use the /h option for the proper commands. thanks!

[MurrayW]

Hi Niels,

 

Thanks for the reply.  I am still probably a month or more away from doing anything, since I am currently using the computer that has the 4852 in it for home automation as well as a shared file server, so I will need to order and build a new HTPC/Automation machine to replace it.  After thinking about it some more this is what I think I will probably do:

 

• Might as well start out with 2TB drives.  Since I have some time, I can probably find a sale on some that are comparable in price/GB to the $70 1TB drives that Newegg has now.  That way, I only have to buy 1/2 as many right now.
  
• Would probably buy 3 2TB drives and start with 2 data + 1 parity + 1 cache (re-use a current drive) connected to the 4 motherboard ports
  
• Keep 4 of my 8 250GB drives connected to the 4852 (or maybe all 8 if I can fit 4 more drives into my case.
  
• That should give me 5TB of space which should last me for a while and when I start running out of room, I can start replacing 250GB drives with 2TB drives, enabling me to eventually get up to ~20TB if I ever need that much space.
  

 

thanks,

Murray

I can't wait any longer to test out unraid on my system using my BC4852 raid card.    I just want to make sure I understand everything correctly to make sure I save all my current data.

 

This is what I currently have:

• Eight 250GB drives connected to the 8 ports of my BC4852 in a traditional Raid5 array.  Currently I have about 500GB filled, but plan on ripping a lot of my DVD's, so I will soon need a bigger array (plus I want to try unraid).
  
• One 200GB OS (XP Pro) drive on one of the motherboard sata ports.  I use this for home automation that I will need to move to a future HTPC build.
  
• One 250GB IDE drive that is empty.
  

 

I think I am going to buy one 2TB drive to use for parity and two 1.5TB drives for data.

 

Migration plans:

• Copy the contents of the current raid array to one of the 1.5TB drives
  
• Install the 2TB drive and the empty 1.5TB drive in place of 2 of the current 250GB drives that are connected to my BC4852.
  
• Create an unraid array that consists of my 2TB drive and one 1.5TB drive from above.
  
• Copy the data from the 1.5TB drive from the first step to the new unraid array.
  
• Verify that all the data is present in the unraid array after the copy
  
• Build my HTPC and move the automation tasks off of the OS drive in my server to the HTPC
  
• Add the 1.5TB drive, 5 of the 250GB drives currently attached to the BC4852, connect the other three 250GB drives currently on my BC4852 to motherboard SATA ports and the 250GB IDE drive to the unraid array.  I might eliminate the 250GB IDE drive if I don't have room for it.  This should give me 5 or 5.25TB of space if I use my IDE drive (2 x 1.5TB + 9 x 0.25TB)
  
• Use the 200GB OS drive as a cache drive
  
• I would end up with 12 or 13 drives
      10 SATA data drives
      1 SATA parity drive
      1 SATA cache drive
      1 IDE data drive (maybe)
  

 

Does that migration plan sound appropriate?

 

One more question.  I have the original Cooler Master CM Stacker case http://www.ocmodshop.com/default.aspx?a=215 and am using three of the 4-in-3 CM modules.  I'd like to move to a true hot-swap stacker to make swapping out drives easier, either 4-in-3 or 5-in-3.  I looked at the wiki for hard disk stackers http://lime-technology.com/wiki/index.php?title=SATA_Hard_Disk_Stackers  Does anyone have any experience with my older version of the CM Stacker that can recommend drive stackers that will work?

 

thanks,

Murray

 

 

 

 

[nia]

Interesting - I feel with you the desire to get things underway :-)

 

First of all, I would definitely keep the parity drive and if at all possible the two 1,5TB drives on the MoBo SATA. Keep in mind that the BC4852 card is on the PCI bus, and that it will be a bottleneck.

 

Then let the smaller drives reside on the BC4852, adding them as you start needing the space. Use the preclear script for easy expansion.

 

I'm a bit unclear as to how many PCs are involved in the migration.

 

I assume you are step-stoning with the temporary move to one of the 1,5TB disks for a reason. Maybe the hardware shuffle dictates this. I assume this to be the case, if you in effect are reusing the same machine that had the BC4852. Otherwise, you could build the unRAID array on a spare harware (the HTPC?) from the start with the 3 new disks. Then migrate them to the final machine. That is similar to what I did here. Works like a charm, and it will allow you to experience first hand the robustness of unRAID in a disaster scenario (simulated burnt MoBo migration) while you still have the data intact on the BC4852 disk set.

 

Another thing: I'm a little puzzled as to why you decide to have a 2TB parity and then two 1,5TB data disks. Is it because you need the rest of the space for some other purpose (if that is even possible - I think not), or do you plan on getting 2 TB's later. Otherwise, 3 like disks either 1,5 or 2 TB seems to make more sense?

 

/Niels

[WeeboTech]

If possible, I would go with using the motherboard ports until exhausted. Especially for the parity drive.

 

If possible, I would consider the largest hard drives within budget/sweetspot that could eliminate spindles to fit within the hardware.

 

I.E. work with as many large spindles that could eliminate the amount of smaller spindles. Then possibly reallocate the smaller ones or sell them off to purchase larger spindles.

 

If you can fit 6 250gb drives on 1 drive, you can save power, maintenance later, sell off the used smaller ones and recoup a few dollars.

 

Just my .02c

 

[MurrayW]

Interesting - I feel with you the desire to get things underway :-)

 

First of all, I would definitely keep the parity drive and if at all possible the two 1,5TB drives on the MoBo SATA. Keep in mind that the BC4852 card is on the PCI bus, and that it will be a bottleneck.

 

Then let the smaller drives reside on the BC4852, adding them as you start needing the space. Use the preclear script for easy expansion.

 

I'm a bit unclear as to how many PCs are involved in the migration.

That makes sense to have the parity drive on one of the motherboard SATA ports and probably my cache drive also.  After that, does it really matter where the other drives are?  And eventually, the 250GB drives will be replaced with larger drives, so I will have to use the BC4852 for some of the larger drives at some point.  And, I really want to test the BC4852 out to make sure it is going to work.

 

There would be 2 PC's.  My HTPC will be in a much smaller case in my entertainment cabinet under my main TV.  My current server (in another room) with the BC4852 in it will become my unraid server.  I haven't even started my HTPC build yet, and yes I know it would make more sense to get that all up and running first before I do anything with my server PC.   But I am just anxious to try out the unraid server and to see if the BC4852 is going to work.  Otherwise, I will have to look at another solution.

 

I assume you are step-stoning with the temporary move to one of the 1,5TB disks for a reason. Maybe the hardware shuffle dictates this. I assume this to be the case, if you in effect are reusing the same machine that had the BC4852. Otherwise, you could build the unRAID array on a spare harware (the HTPC?) from the start with the 3 new disks. Then migrate them to the final machine. That is similar to what I did here. Works like a charm, and it will allow you to experience first hand the robustness of unRAID in a disaster scenario (simulated burnt MoBo migration) while you still have the data intact on the BC4852 disk set.

The reason for putting my current data on one of the 1.5TB drives is to be able to take 1 or more drives off my BC4852 and not lose the data until I have had a chance to copy it to the unraid array.

 

Another thing: I'm a little puzzled as to why you decide to have a 2TB parity and then two 1,5TB data disks. Is it because you need the rest of the space for some other purpose (if that is even possible - I think not), or do you plan on getting 2 TB's later. Otherwise, 3 like disks either 1,5 or 2 TB seems to make more sense?

 

/Niels

The only reason for going with different sized drives is cost.  I can get 1.5TB drives for $110 and the 2TB for $160.  I want to start with a larger parity drive because I am sure that sometime in the future I will be using 2TB drives for replacements.

 

thanks for your help,

Murray

[MurrayW]

If possible, I would go with using the motherboard ports until exhausted. Especially for the parity drive.

 

If possible, I would consider the largest hard drives within budget/sweetspot that could eliminate spindles to fit within the hardware.

 

I.E. work with as many large spindles that could eliminate the amount of smaller spindles. Then possibly reallocate the smaller ones or sell them off to purchase larger spindles.

 

If you can fit 6 250gb drives on 1 drive, you can save power, maintenance later, sell off the used smaller ones and recoup a few dollars.

 

Just my .02c

 

Yeah, I've gone back and forth on wondering if it is even worth starting with the 250GB drives or just taking them out from the start.  Something about already having them and not having a replacement makes it hard for me to get rid of them!  For a little over $100, I could replace the capacity from the start.  Saving maintenance later is a BIG selling point for me...that is why I am considering replacing my hard drive stackers, even though I could use that same money to buy a couple more hard drives.

 

As I mentioned in my reply to nia, one of the driving factors for using the BC4852 is just to make sure it is going to work.  If my parity and cache drives are on motherboard SATA ports, would I ever notice the slower BC4852 ports for some of my data drives?

 

thanks,

Murray

[WeeboTech]

As I mentioned in my reply to nia, one of the driving factors for using the BC4852 is just to make sure it is going to work.  If my parity and cache drives are on motherboard SATA ports, would I ever notice the slower BC4852 ports for some of my data drives?

 

Believe me, I want to know if the controller works too!! It would be a real help to the community.

I would go with the denser drives, Then experiment on the smaller 250gb drives and BC4852 card after I was confident the system was stable and of adequate performance.

 

As far as noticing the slower bus speed on the BC4852, probably not for single drive access.

If you use the larger drives on the motherboard and the array of smaller drives on the BC4852, you would notice the slower speed during parity generation/parity check for the amount of space of the drives on the BC4852. I.E. once parity generation/check passes all the drives on the controller, the speed will pickup to what the system & bus can handle.