Proxmox Fiber channel storage with MPIO Setting

[simonmok113]

Hello guys , i have a problem with Fiber channel storage multipath setting .
I have a 3 pve node in cluster. I add the Fiber channel storage (SAN) as share storage with (Share-LVM) for live migration and High availability and I setup 2 LUN with both node . Those LUN are different Tiers storage with SSD and HDD.
The Storage LUN can mount to both node. However, the multipath (MPIO) cannot success to run round robin load balancing for no down time failover with Fiber channel storage and SAN switch. And now, It only run as active / standby (active ready running / active ghost running) , So if run no down time failover with SAN normally is (active ready running / active ready running). How to solve this issues, the configuration file in the attachment and below.

WWID information in the below:
HDD Tier: 36e0cc7a10010a94c0008fad000000002
SSD Tier: 36e0cc7a10010a94c1213fed600000004

The /etc/multipath.conf in the below:
root@TKO01:~# cat /etc/multipath.conf
defaults {
polling_interval 2
path_selector "round-robin 0"
path_grouping_policy multibus
uid_attribute ID_SERIAL
rr_min_io 100
failback immediate
no_path_retry queue
user_friendly_names yes
}

blacklist {
wwid .*
}

blacklist_exceptions {
wwid "36e0cc7a10010a94c0008fad000000002"
wwid "36e0cc7a10010a94c1213fed600000004"
}
multipaths {
multipath {
wwid "36e0cc7a10010a94c0008fad000000002"
alias mpath000
}
multipath {
wwid "36e0cc7a10010a94c1213fed600000004"
alias mpath001
}
}

Hardware details:
SAN Storage model: Huawei Oceanstor 5500 V5
HBA Card: Emulex,FC HBA,16Gb(LPe16002B),2-Port,SFP+

Thanks

 

[LnxBil]

Enable ALUA on the storage side. This is not a client issue, but a server issue. Some storages simply don't support active/active.

 

[simonmok113]

Enable ALUA on the storage side. This is not a client issue, but a server issue. Some storages simply don't support active/active.

Hello LnxBil, Thanks for your reply.
Before i communicate with huawei , they ask the storage support active/active mpio since they install multipath tools in centos or redhat or esxi can success to run active/active, It seems like mpio or os based problem.
Ummm.... This is a difficult problem since less cases , i am curious if run dell emc storage the multipath.conf is look like ?
Thanks

 

[LnxBil]

We are maintaining a huawei xsg1 full nvme storage with multipath and alua and it was straight-forward to set up as any other storage I worked on in the past decade. We do not have any additional configuration settings. Often, vendors provide their own multipath.conf settings, maybe you can ask for the specific settings?

 

[simonmok113]

We are maintaining a huawei xsg1 full nvme storage with multipath and alua and it was straight-forward to set up as any other storage I worked on in the past decade. We do not have any additional configuration settings. Often, vendors provide their own multipath.conf settings, maybe you can ask for the specific settings?

Huawei they provide the multipath tool to us. However, it cannot support debian only support on redhat / centos / suse linux.
They provide multipath tools call Ultrapath.....

 

[LnxBil]

Huawei they provide the multipath tool to us. However, it cannot support debian only support on redhat / centos / suse linux.
They provide multipath tools call Ultrapath.....

We don't use that, just the plain OS stuff.

 

[simonmok113]

We don't use that, just the plain OS stuff.

LnxBil could you places your multipath.conf for reference?
Since i use proxmox default multipath tools but still active ghost running.
Thanks

 

[LnxBil]

LnxBil could you places your multipath.conf for reference?

As I wrote in comment 4, I do not have any configuration present besides the naming of the devices.

For older EMC DGC devices, we needed to force alua, maybe that could help for you too:

devices {
  device {
         vendor "DGC"
         product "*"
         hardware_handler "1 alua"
         prio "alua"
        }
}

You have to adapt the vendor and product of course.