upgrading old freenas

[scanray]

Hi,

I received an old computer to retrieve information freenas because the discs were spoiled. This is ancient history. I could already recover most information.The hardware is as follows:

Intel(R) Xeon(R) CPU E3110 @ 3.00GHz
Intel Corporation S3210SH
2GB ram
1 HD 500GB (system)
Board 9560SE (16 HD)
16 x 2TB HD Hitashi
NORCO RPC-3216 Black 3U Rackmount Server Chassis

I've used raid5 on a server for more than 10 years ago. In linux.
I've always used for disks raid cards, but only as an interface for AHCI ports and I created the raid by the software.I've never had problems with the raid.
But, since I'm reading information with freenas, I have entered some doubts with raid5.

Clearly disks card I have in the freenas is ancient and at any time could have a failure, which would be serious because it would be very difficult to get another one.
So you could recommend me the best way to take advantage of this hardware with installation of freenas?
- I had thought to use the RAID card only as ahci interface and create the raid disks for software, but it is really advisable?
- It would be necessary to increase an SSD as cache? (For the zfs)

The equipment is to be used only as fileserve/backup. It is for a house making videos.

Any comments/help is appreciated
thanks

 

[cyberjock]

Read our hardware recommendations.. you have 1/4 of the minimum RAM for FreeNAS. Unless you plan to ditch the RAID card and spend money on RAM and a good HBA, better find another OS to use.

As old as that hardware is, it's basically inadequate in all of the major ways for FreeNAS. Your CPU has a FSB, which is going to be a nasty bottleneck for ZFS.... etc etc etc.

 

[scanray]

Hi cyberjock

According to your comments, I should change MB, CPU, RAM and HBA. In other words, buy a new server.
Well, anyway I buy memory. I think this MB only supports 16GB.
Server owners will not spend much money.
I already told them to buy 16 x 4TB HD and were frightened of the cost (strangely the 16HD were damaged. 4 with DCB Read Failure and the other 20% of remaining life)

I know that the equipment is old, but is connected to a network 1GB and will only be used for backup. The maximum transfer rate is 125MB/s. If this server can give me that increasing the memory, the owners will be happy.

Is posible to use another file system in freenas? Or just use ZFS?
In any case, if I can not use ZFS for equipment conditions, which another you recommend me to use? ext4, UFS, Reiser4, btrfs ....

thanks for your reply

 

[jgreco]

Note that the 3Ware 9650 does not support drives larger than 2TB except on special firmware (check with Avago support on the current status, might be released now).

It sounds like several of the drives failed ("DCB Read Failure" means that the 3Ware Disk Control Block was unretrievable, and since that's stored redundantly, suggests large scale disk failure). It isn't that strange for disks to fail periodically, and if you aren't monitoring for that issue and replacing them as they fail, sooner or later your storage WILL fail.

You might want to check the reported peak temperatures of each disk in that Norco chassis. The Norcos are notorious for less-than-ideal airflow, and when drives run too warm, they tend to fail more frequently. SMART will have that data if you query the drives. If it is reporting that the drives are overly warm, this should probably be addressed, either through remediation or chassis replacement.

A quality FreeNAS install doesn't wait for a hardware RAID controller to decide that a disk is "dead" - we like to look for problems proactively, before they progress to a data-impacting or disk-failure event. Reading in between the lines, it sounds kind of like the owners set the box in a corner and didn't pay any attention to problems that were developing. That's always going to lead to failures. If they're interested in a more reliable storage platform, they're encouraged to follow the hardware recommendations and build a system suitable for FreeNAS, and set it up with SMART monitoring, email alerts, and all the other fun stuff, and it will serve them well. If they're not interested in a more reliable storage platform, suggest loading something like Openfiler on it, sticking it back in the corner, and then waiting for the next failure.

 

[scanray]

Hi @jgreco

Note that the 3Ware 9650 does not support drives larger than 2TB except on special firmware (check with Avago support on the current status, might be released now).

I've already contacted Avago. I already have the firmware to update. I'm hoping they get the discs to make the upgrade.

It sounds like several of the drives failed ("DCB Read Failure" means that the 3Ware Disk Control Block was unretrievable, and since that's stored redundantly, suggests large scale disk failure). It isn't that strange for disks to fail periodically, and if you aren't monitoring for that issue and replacing them as they fail, sooner or later your storage WILL fail.

Unfortunately I do not know who bought this server. I do not know who set up the server. I received spoiled. In the configuration it was disabled smart. Obviously they had not configured the system to do the reports.

You might want to check the reported peak temperatures of each disk in that Norco chassis. The Norcos are notorious for less-than-ideal airflow, and when drives run too warm, they tend to fail more frequently. SMART will have that data if you query the drives. If it is reporting that the drives are overly warm, this should probably be addressed, either through remediation or chassis replacement.

When I was recovering the information, I have noticed the increase in temperature in both the discs and the cpu. This seemed too strange, especially since it was a simple copy. It is also possible overwork due to damaged disks.
Yes, definitely I have decided to change the case. I hope that the owners accept the change.

A quality FreeNAS install doesn't wait for a hardware RAID controller to decide that a disk is "dead" - we like to look for problems proactively, before they progress to a data-impacting or disk-failure event. Reading in between the lines, it sounds kind of like the owners set the box in a corner and didn't pay any attention to problems that were developing. That's always going to lead to failures. If they're interested in a more reliable storage platform, they're encouraged to follow the hardware recommendations and build a system suitable for FreeNAS, and set it up with SMART monitoring, email alerts, and all the other fun stuff, and it will serve them well. If they're not interested in a more reliable storage platform, suggest loading something like Openfiler on it, sticking it back in the corner, and then waiting for the next failure.

Yes, I think they bought the server because they hear that this was the solution to store all the information. I think not well advised.
I passed them a budget to buy a new server. Because they will buy all disks, the difference to the server is not much.
This old server, with disks that have new (10), I'm going to put a Linux or Openfiler (which I have not ever used) and that it turned on only when they want to recover some old information

Thank you very much @jgreco, @cyberjock, for the help, has been enlightening

 

[jgreco]

Well, you've been pointed in a better direction. Here's hoping that you get to go.