Hey!
First of all thank you very much for developing linstor-gateway. It's a very handy software indeed and a great addition to the Linstor stack.
I came across a limition and can think of a rather easy way to fix it.
I am running a zfs pool and therefore am using the linstor zfs storage pool backend.
I created a highly available NFS export by using linstor-gateway nfs create [...] --filesystem ext4.
This will automatically create a new zvol and format it as ext4 (after putting drbd in between).
The problem is that as of now zvols default to a volblocksize of 16k and can't go lower than 8k (=minimum value).
This does not align with the default block size of ext4 that get's used by the filesystem created bylinstor-gateway which in turn results in the performance being rather poor compared to a manually created zvol --> ext4 --> nfs setup.
It would be awesome if "linstor-gateway nfs create" had another flag that would offer the option to pass flags to the mkfs.ext4 / mkfs.xfs invocations.
That way people have a tool for optimizing the storage stack in order to get the most out of the hardware.
If there already is a way to configure ext4 / xfs block-/fragment-size elsewhere in the linstor stack which I could use as an alternative, please let me know. Otherwise I would be really greatful if the flag described above would get added.
Thank you in advance!
Best regards
Hey!
First of all thank you very much for developing linstor-gateway. It's a very handy software indeed and a great addition to the Linstor stack.
I came across a limition and can think of a rather easy way to fix it.
I am running a zfs pool and therefore am using the linstor zfs storage pool backend.
I created a highly available NFS export by using linstor-gateway nfs create [...] --filesystem ext4.
This will automatically create a new zvol and format it as ext4 (after putting drbd in between).
The problem is that as of now zvols default to a volblocksize of 16k and can't go lower than 8k (=minimum value).
This does not align with the default block size of ext4 that get's used by the filesystem created bylinstor-gateway which in turn results in the performance being rather poor compared to a manually created zvol --> ext4 --> nfs setup.
It would be awesome if "linstor-gateway nfs create" had another flag that would offer the option to pass flags to the mkfs.ext4 / mkfs.xfs invocations.
That way people have a tool for optimizing the storage stack in order to get the most out of the hardware.
If there already is a way to configure ext4 / xfs block-/fragment-size elsewhere in the linstor stack which I could use as an alternative, please let me know. Otherwise I would be really greatful if the flag described above would get added.
Thank you in advance!
Best regards