#E2800A

1 messages · Page 1 of 1 (latest)

red nova
#

I have a two contoller E2800A-32GB .

using it with proxmox v8 Linux 6.14.11-7-bpo12-pve

Getting random read errors via mulitpath such as

May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdb, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdb, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sde, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sde, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdh, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdh, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdg, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdg, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdk, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2
May 04 10:47:48 proxmox0105 kernel: I/O error, dev sdk, sector 0 op 0x1:(WRITE) flags 0x800 phys_seg 0 prio class 2

At this point I think there is a version of multipath that might fix the problem but can't get to it on this proxmox release

So finally disabled autobalance to stop the multipath shifting all the time.

So starting to wonder if in the beginning the netapp applicance was targetd at the correct linux type Linux (MPP/RDAC)

If not, what is best way to change, as I think I picked up that changes would only affect new, and not existing?

bit more info
Firmware version:
08.81.00.03
NVSRAM version:
N280X-881834-D01

Just a bit more history, we have had this unit since new, and it is still under a contract. But have concern that proxmox kernel updates might be issues

opaque laurel
#

To be clear, I'm not sure this is a tested and supported configuration (e-series). The only interop testing we've done to-date is with ONTAP on our FAS/AFF systems. To my knowledge, we've not put the eSeries platform through its paces with regards to proxmox.

#

Is this is a homelab project, jump down to #┊・homelabs channel, but if you're trying to run this as a production hypervisor workload, there's some considerations there.

red nova
#

no homelab, its commerial about 60 VMs on it now

#

my research has lead me to same

opaque laurel
#

OK, let me check with the eSeries team and point them in this direction for ya

red nova
#

thank you sir!!

red nova
red nova
#

※ recap: Cleaning up ghost path sector 0 WRITE errors from the NetApp array. Errors are still firing
every 5 minutes from the Linux rdac hardware handler — not fixable from the NetApp side. Next
step is researching whether changing SANtricity host type to LnxDHALUA plus switching multipath to
ALUA would eliminate them.

quasi sandal
#

not sure about proxmox, but with Linux and mpath, you should not use the direct paths like /dev/sda... rather use the /dev/mapper/device... with the E-Series the LUN is either on controller 1 or 2.. but can be accessed via both controllers.. but ALUA should make sure the best path is selected... I would think proxmox is capable of all this, but we only use NFS for proxmox...