What comes out of the box with RHEL 8 - XFS
You do not need to worry about what file systems is used in the StoreOnce appliance, HPE has already decided that for you.
Hi Nicolai, thank you for your response, but the question was focused to the media server (with Linux) that we need install and configure, the StoreOnce has its own operating system and it don't enter in this discussion.
This new linux media server will have RHEL 8.3 and we must install the HPE NBU OST plugin to integrate with de StoreOnce appliance, so, the question was what is the recommended filesystem structure for the linux RHEL media server, for example:
Create various mount points under LVM physical volumen(pv)/vg/lv:
Thank you for your help.
200 G for a system that only needs CPU and memory are to much.
Nicolai is right. There is no need to spend disk for a media server that his only job is to deduplicate and feed data to the storeonce.
My opinion is to configure 100G (and it is too much to) for root FS and install netbackup there. Or do whatever you like. There is no difference.
my recommendations are:
For the file system as already stated use the native XFS.
For the layout, use whatever you are most conformatble with. I'd personally go with LVM as it has the advantage of the ability to change it if you find you need more space later (more easily than running the file system on the raw partition).
For disk space and other server recommendations go with what others have said (although I'm curious as to why @StefanosM wants to avoid FC connectivity to the StoreOnce - I have a couple of customers that use this without issue, can you elaborate please?).
While 200 GB might be a bit much as has been mentioned by others, definitely give yourself some elbow room and don't trim it down too much.
Maybe 50 GB for / and another 50-75 GB for /usr/openv/ . Remember it's not just about storage for the binaries and such, at some point in the future you're going to want to crank logging up all the way, maybe for several days running (long weekend?), to troubleshoot some problem. Depending on the issue that could generate some decent-sized NBU logs and it'd be nice not to fill up the entire filesystem with them & crash your other backups. =)
Likewise, core dumps are useful troubleshooting tools on occasion and it's nice to have room to generate them.
Food for thought.
I second the request for @StefanosM to elaborate on not using FC connection to StoreOnce.
Also, as @davidmoline say, stay with XFS - with default file system for RHEL8 supported by NetBackup, there is one less change to worry about during a disaster/rebuild.
Capacity of various mountpoints - ask for recommendations from your Linux sysadmins, then use your own requirements. E.g. which partitions do you want to be separately mounted, space required for OS in /, how much logging space are you comfortable with, etc. LVM is easiest way to go if you don't have any other option as it can be expanded easily. I have a storage foundations volume presented to the media server for the NetBackup installation (100GB from SAN) and the rest of the partitions are LVM (local disk).
Red Hat has different file system layout depending on the hardening requirements. Use LVM for volume management so it's easier to extend file systems. Do go with "hole filesystem layouts".
For /usr/openv I would go with a 60 GB Volume
For /usr/openv/netbackup/logs and /usr/openv/logs I would go for around 20 GB each to hold debug logs (if enabled).