Partition on large iscsi target failed

[ccnet]

Hi
Setup
Proxmox backup server 4.2 on dedicated hardware (24 x Intel(R) Xeon(R) CPU E5-2640 0 @ 2.50GHz, Ram 72 Go, Local Storage 1,45 To)
iSCSI storage on dedicated Synology (13To)
iscsiadm -m session return :
tcp: [1] 192.168.10.113:3260,1 iqn.2000-01.com.synology:SAN-03.Pbs-00.726590134d2 (non-flash)
tcp: [2] 192.168.10.113:3260,1 iqn.2000-01.com.synology:SAN-03.default-target.726590134d2 (non-flash)

First : 600 Go
Second : 13,7 To

lsblk show :
sdb 8:16 0 13.7T 0 disk
sdc 8:32 0 600G 0 disk /mnt/datastore/san-03

Issue
mkfs.ext4 /dev/sdc is working and i can create a datastore.
mkfs.ext4 /dev/sdb or mkfs.zfs /dev/sdb seems never end after hours. And finaly /dev/sdb can't be used.

dmesg show various errors :
[Aug23 14:56] perf: interrupt took too long (2504 > 2500), lowering kernel.perf_event_max_sample_rate to 79000
[Aug23 15:56] EXT4-fs (sdb): VFS: Can't find ext4 filesystem
[ +0.001022] EXT4-fs (sdb): VFS: Can't find ext4 filesystem
[ +0.000485] EXT4-fs (sdb): VFS: Can't find ext4 filesystem
[ +0.000724] FAT-fs (sdb): bogus number of reserved sectors
[ +0.000004] FAT-fs (sdb): Can't find a valid FAT filesystem

Thank you for reading

 

[UdoB]

iSCSI storage on dedicated Synology (13To)

What kind of disks are used and how many. How is the storage organized?

You can (just for debugging/testing) try to use a much, much smaller target. If that works you know the mechanism is ok and the capacity raises problems.

----
PBS needs IOPS. Compare https://pbs.proxmox.com/docs/installation.html#system-requirements

If you have only rotating rust, no fast "official" write-caching device in the Syno and organize it at SHR2 (or what the RaidZ2-equivalent is called) and you have a slow CPU then the construct may just be too slow to work reliably. I've been there, several years ago, using NFS though. At the beginning (with an empty store) it worked, but later on I was not able to list the content of the datastore, instead I just got timeouts...

But nevertheless: ymmv!

 

[ccnet]

That's classical hdd. Four disks managing by a raid hardware controler.
i know the mechanism is ok.
/dev/sdc is a smaller lun (600 Go) on the same storage device. It's working but/dev/sdb, a large lun of 13To, is not working.
Ext4 as xfs can handle large system files. So why the size is a problem ?