[SOLVED] issue with ZFS on HP smart arraw p420i

killmasta93

Renowned Member
Aug 13, 2017
966
56
68
30
Hi,
I was wondering if someone else has had the same issue? I went to the bios configured it AHCI SATA but for some odd reason it wont get the disk for ZFS.

Thank you
 
Hey!

If i understand correctly, you've set the AHCI SATA, but this does not necessarily influence the drives. First of all - are the drives SATA? Secondly - does p420i allow for the drive pass-through?

I assume that p420i is not really designed for pass-trough mode. Here's a link that is a complicated guide on how to achieve this (potentially resulting in unstable system).

As a workaround you can create a separate RAID0 array per each drive. This way the ZFS will see the drives. It results in a realatively unstable system though, so i suggest you look for a pass-through controller.
 
AFAIK the P420i is an RAID Controller, they have his own BIOS. Normally you can not configure AHCI in the Controller itself, the Configuations are only applie for the onboard SATA Ports. I would recommended to buy an HBA, an other but not really good way is to configure an own VD as RAID0 for each disk.
 
Thanks for the reply, so better to buy what @Vladimir Bulgaru suggested?
forgot to mention DL380P HP it also has SSD would that matter?
There may be different HBA controllers out there. I am recommending this one since i have it in my DL360p and i know for a fact that it works. You will need to get new cables, though, as the ones that currently co into the motherboard won't fit into the 2 ports on the HBA, so it's best to get one with the cables.
 
  • Like
Reactions: killmasta93
Do not use this controller for ZFS, it will damage your system, sooner or later with positivity. I already have enough experience.
Hello @fireon , could you share your problems with this controller ?

i have HP DL380e Gen8 with P420 and did thoroughly test it in passtrough mode with ZFS and 12 SATA disks , i.e. lots of IOPs, bandwidth saturation with dd read/write of uncomressed data streams, scrub etc - i cannot see any problem, it works rock solid for me.

getting about 2GB/s read performance when doing scrub, which is quite good, imho.

some guy named gmc_5303 at
https://www.reddit.com/r/Proxmox/co.../?utm_source=share&utm_medium=web2x&context=3 telling the same
 
Last edited:
I had this on two DL120s, and one DL370. I don't know the controller type anymore. I only know that I called HP support and they confirmed that these controllers responded as such in HBA mode. And yes, for a while they did. At some point there was an I/O loop and after a reboot which became a reset, that was it. With one DL120 we could still save the data, there only the machine didn't start anymore, with the other two it was definitely over.

And also the Proxmox support confirmed that HP does not use real HBA controllers but the Raidfirmware only does HBA. A reflash fixes the issue of course, but without warranty from HP.

Therefore we only have Thomas Krenn servers for ZFS.
 
>I don't know the controller type anymore

so please do not generalize then.

https://forums.servethehome.com/index.php?threads/p420i-and-ssds.21817/post-227571

arglebargle - May 13, 2019
I've been playing with a DL380p gen8 this week and I can add a bit to this. With current firmware I've had zero issues with the p420i in hbamode. Everything works, SMART pass through is flawless on both SATA and SAS drives and I've had no issues with dropped drives while running badblocks against 6 5tb SATA drives for 4 days and simultaneously slamming the two SSDs with max IOPS fio tests. I've run fio simultaneously on up to six separate SSDs (2 Samsung 850 pro, 2 SAS SanDisk Optimus and 2 STec S840s) without any hiccups at all.

Just wanted to throw it out there that the p420i at least is a totally viable HBA.

Totally stock setup here: onboard p420i, stock SAS cables to the 8 drive cage up front. I'm using whatever the cheapest carriers I could find on eBay are, I think they're legit HP.
 
  • Like
Reactions: ianmbetts
meanwhile i want to provide an update on this, i i did a citation "it would be totally viable". that's not true.

>i have HP DL380e Gen8 with P420 and did thoroughly test it in passtrough mode with ZFS and 12 SATA disks , i.e.
>lots of IOPs, bandwidth saturation with dd read/write of uncomressed data streams, scrub etc - i cannot see any
>problem, it works rock solid for me.

>Just wanted to throw it out there that the p420i at least is a totally viable HBA.


it seems to really depend which disk you use and what type of errors occur.

based on recent experiences with my p420i in hba mode while thoroughly testing/wiping older harddisks and sorting out old/bad disks for trash , i can tell this hba (compared to lsi) really sucks regarding error handling. while doing nothing but "dd" to or from several disks concurrently , i had multiple hangs of the whole controller and i/o stopped completely to all disks.

that would explain why i had at least two or three complete lockups with zfs when using that controller in a test server a while ago.

also see https://community.hpe.com/t5/ProLia...-Gen8-P420i-Controller-Lock-Up-s/td-p/7059791

conclusion:

p420i in hba mode - use at your own risk and don't use it in production. test thoroughly and test with bad disks.
 
Last edited:
I have a ProLiant ML310e Gen8, which controller should I use for zfs? (VM trueNAS)

i use now AHCI mode
internal controller Smart Array B120i only supports

Code:
x2 6Gb/s SATA physical links (compatible with 3Gb/s SATA )NOTE: Two SATA ports will be running at 6Gb/s and reaming ports willbe running at 3Gb/s

how to expand the configuration?
 
Last edited:
meanwhile i want to provide an update on this, i i did a citation "it would be totally viable". that's not true.

>i have HP DL380e Gen8 with P420 and did thoroughly test it in passtrough mode with ZFS and 12 SATA disks , i.e.
>lots of IOPs, bandwidth saturation with dd read/write of uncomressed data streams, scrub etc - i cannot see any
>problem, it works rock solid for me.

>Just wanted to throw it out there that the p420i at least is a totally viable HBA.


it seems to really depend which disk you use and what type of errors occur.

based on recent experiences with my p420i in hba mode while thoroughly testing/wiping older harddisks and sorting out old/bad disks for trash , i can tell this hba (compared to lsi) really sucks regarding error handling. while doing nothing but "dd" to or from several disks concurrently , i had multiple hangs of the whole controller and i/o stopped completely to all disks.

that would explain why i had at least two or three complete lockups with zfs when using that controller in a test server a while ago.

also see https://community.hpe.com/t5/ProLia...-Gen8-P420i-Controller-Lock-Up-s/td-p/7059791

conclusion:

p420i in hba mode - use at your own risk and don't use it in production. test thoroughly and test with bad disks.
RolandK

As you have done some testing with the p420I on the gen8.

question: to run proxmox on G8 server 64 G ram (ecc) with the above controller boot disk will be ssd. not planning to use ZFS at this time is it required to run in HBA mode?
suggested configuration would be appreciated.
right now the system sees the boot drive separtately from the 6 other 2.5 hp 600 gb drive.

boot drive is crucial 500G ssd. the rest are

Toshiba MBF2600RC - Hard Drive - 600 GB - SAS (CR5527)​


thank you sir
richs
 
meanwhile i want to provide an update on this, i i did a citation "it would be totally viable". that's not true.

>i have HP DL380e Gen8 with P420 and did thoroughly test it in passtrough mode with ZFS and 12 SATA disks , i.e.
>lots of IOPs, bandwidth saturation with dd read/write of uncomressed data streams, scrub etc - i cannot see any
>problem, it works rock solid for me.

>Just wanted to throw it out there that the p420i at least is a totally viable HBA.


it seems to really depend which disk you use and what type of errors occur.

based on recent experiences with my p420i in hba mode while thoroughly testing/wiping older harddisks and sorting out old/bad disks for trash , i can tell this hba (compared to lsi) really sucks regarding error handling. while doing nothing but "dd" to or from several disks concurrently , i had multiple hangs of the whole controller and i/o stopped completely to all disks.

that would explain why i had at least two or three complete lockups with zfs when using that controller in a test server a while ago.

also see https://community.hpe.com/t5/ProLia...-Gen8-P420i-Controller-Lock-Up-s/td-p/7059791

conclusion:

p420i in hba mode - use at your own risk and don't use it in production. test thoroughly and test with bad disks.
I have a a few of these dl380p, currently running a ceph cluster with osds on raid0 and very happy with it. I am about to build a new cluster and was planning to use zfs replication rather than ceph. I was encouraged to read your first post and less so your second post.

One question in your tests did have the hpsa mode driver explicitly enabled in the kernel ? (I have read that the kernel will prefer the legacy cciss driver unless told otherwise)
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!