Replaced a fail disk in RAID-Z1?
Actually, this is what i was thinking and it could be the best possible thingy. Backup the whole important data somewhere else safely and then replace the disk!
So, either i do resilvering or backup the data elsewhere, it will have the same stress on the disk as in both the case, the heads and platter will move but resilvering will be done at faster speeds so there is less chance for a second drive failure. Is that what you mean?
Can you explain how do you keep an exact duplicate HDD in cold storage? And what do you mean by cold storage here? I guess probably by removing the drive X, keep it spare and then add the disk Y and resilver the pool and continue with the NAS. When a disk fails, remove it and replace it with disk X again and the pool is back to the healthy state. Am i thinking correct?
When I configure a server, I buy at least one extra identical drive, and put that on a shelf. If a drive in the array fails, I pull that dead one out of the server, I pull out the brand new drive, unbox it, stick it in the slot that the bad one just came out of, and re-silver. Then I send the bad drive back to the manufacture, and hope and pray that they send me a new one, rather than some BS “refurbished“ one…and ideally just order a new NEW drive for cold storage.
Okay cool. Thanks for the info!
Potentially / theoretically.
Basically the longer you wait, (it’s been 7 days since you started this forum thread), the higher the risk of another failure. (Well, to be fair, 7 days is nothing. 7 weeks would be something…)
So, do one or the other. In my opinion, doing something is better than waiting and getting the supposed perfect answer.
Unless you get a second completely failed disk, bad blocks only affect a limited set, (like a single file). And if the bad block is in Metadata, (regular has 2 copies even on top of RAID-Z1, critical has 3 copies…), you have a good chance it won’t be impacting at all.
2 Likes
I’m waiting for the drive to arrive 
Yea, yeah.
Cool. If i recall it correctly, i’ve resilvered it like 4-8 times in total before this failure in the past three years i guess. So, yeah, will try to resilver.
Thank you for all the guidance and help!
1 Like
I’m running a HP DL380eG8 with six 3T SAS drives. The controller shown in the photo is an HPE Dynamic Smart Array B120i/B320i SAS RAID Controller. I use it because there is no RAID setup to do in hardware, and adding new drives or replacing drives that are failing doesn’t require any tricks. On my old faithful TrueNAS Core I get 120 MB/sec transfer rate using Filezilla when uploading fresh content. When I built the TrueNAS Scale 25.04.1 system… populating the content was done using either Filezilla or SMB. Both limited the transfer rate to 11 MB/set. My home network is Cisco Gigabit switch and has always had fantastic transfer rates… in some cases approaching 400 MB/sec. So my question is… using similar hardware why doesn’t Scale go any faster than 11 MB/sec // The controller shows up at boot time as an H200. HPE says no drivers are required when using this in a DL-380E. Which it is made for.
I’d suggest that new problems always get new threads with the unique things identified in the title.
In this case… its Low bandwidth with HPE Dynamic SMART ARRAY RAID Controller.
FYI… I don’t know of anyone that tests with that. So it is a mystery.
The same software gets 2.8GB/s on high performance hardware. Clearly there is something wrong. I’d do independent tests of SMB, ZFS and network. I don’t know whether the client is vanilla or chocolate… against better discussion for a new thread.
Redhat and others calim that a driver is needed for each OS.
Is the HP Smart Array B320i, B140i, B120i, B110i controller supported by RHEL or RHELOSP? - Red Hat Customer Portal.
Changing from CORE to SCALE means a new driver… I doubt that SCALE has it.
1 Like
I may have to pull it out and try configuring the 420i controller for pass through.
That would work.
You could confirm the theory with a local test of ZFS performance 1st. If local performance is bad, then we know its not network or SMB.
Definately NOT a NETWORK or SMB problem. The B420i controller works great on Windoz or FreeBSD. But needs a driver for Linux. I’ve copied 4TB onto the pool, and am not ready to do something that would loose it. Switching to the 420i embedded controller would require me to reconfigure the controllrr to PASS THROUHH. And there is no guarantee it would use the pool written in software by Tnas SCALE.
it ain’t no big thing, just days of SMB copy time… Everything is on another server that is working great.
Kinda like that Tnas is using Docker now. But in the final reel, it has to work with very little added expense, or its not gonna happen.
Great… at least we know where the issue is.
I asssume debian linux does not have the driver either?
Yes. I never gave it any worry because it worked great with FreeBSD in TrueNAS Core. I will likely stay with Core as there is no reason to change or update.
1 Like
After you suggested that Linux didn’t support the B320i HBA adapter HP 684896-001 I decided to dump the TrueNAS Scale 25.04.1 and boot up Proxmox, which is also a Linux based system. After loading a Mint 21.3 OS I transfered some large files and got in excess of 130 MB/sec. So your idea that Scale didn’t perform because of being Linux is unlikely. Now it could be that Scale needs to add drivers for such.
For the time being I’m staying with Proxmox. It is a sensible system with few limitations.
My comment was that it appeared that a driver might be needed… it would be useful to know if it is.
1 Like
None found from HP