Mounting logical drives on server

[billtekker]

Hi,

I have a Raspberry Pi Debian GNU/Linux 11 (bullseye) operating system installed on the HPE Proliant ML350 G10 server. It has a 240gb HPE Logical Volume mounted at Filesystem Root (Ext4) but I'm running out of space on that and want to make use of the 2.4tb HPE Logical volume that's there, however it is unmounted. My question is, if I mount the 2.4tb logical volume (which consists of 2 x HPE - 1.2TB - SAS 12Gb/s - 10K - HDD 2.5" drives), is there any risk of wiping any of the data or file structure on the 239gb logical volume that's mounted in root?

P.S. The 240gb logical volume consists of an HPE - 240GB - SATA 6Gb/s - SSD 2.5" drive. There is another HPE - 240GB - SATA 6Gb/s - SSD 2.5" in the rack but for some reason it is not being recognised.

Thanks in advance.

[swampdog]

There's not enough info for definitive answer. It may be that "HPE Logical Volume" is in fact LVM2 in which case the answer is "probably not" with the caveat you need know about LVM2 to find the answer for sure plus with it being a server there's likely so be some kind of raid going on so you'd need to know about that as well.

For instance, on my pair of nas..

Code: Select all

foo@sdhp0:~$ sudo lvs | grep ALL
  isALL           sdhp0_data -wi-ao----  16.00g 
  
foo@sdhp1:~$ sudo lvs | grep ALL
  isALL           sdhp1_data -wi-ao----   16.00g 

..it appears to first sight nothing uses the above. Neither are mounted and neither are formatted (at this level). In fact what they are is a pair of iscsi devices which an initiator typically sees as two disks which could be anything (as I use them for quick experiments).

Thus, unless you configured the server, assume you will break something badly.

[billtekker]

Thanks for the response swampdog. I didn't configure the server and have no access to the person who did so I'm not sure how it was configured. None of the commands to show the logical volumes or physical discs seem to work (such as sudo lvdisplay or pvdisplay) but it says that lvm2 is installed (see attached screenshot).

IMG_20230602_084903.jpg (141.51 KiB) Viewed 1132 times

Is there any way of accessing LVM2, in order to see the configuration and see how it manages the logical volumes and physical volumes etc? I've seen the following: https://wiki.debian.org/LVM#Installation

"system-config-lvm" - it's a utility for graphically configuring Logical Volumes. Have you any insight into whether or not it's worth installing that?

Cheers.

[swampdog]

Sorry for delay. Not getting much computer time atm.

Is that the rpi in the above screenshot? The HP server is intel based so raspOS cannot be running *on* it so now I'm confused. If you type "sudo pvs" on the HP server then if there is anything LVM based configured then you will see a result. Example:

Code: Select all

foo@sdhp0:~$ sudo pvs
  PV         VG         Fmt  Attr PSize   PFree   
  /dev/md2   sdhp0_data lvm2 a--    7.21t <108.54g
  /dev/sda2  sdhp0      lvm2 a--  <63.07g  <48.85g
  /dev/sdb1  sdhp0_owl  lvm2 a--  508.00m       0 
  /dev/sdb2  timeshift  lvm2 a--   32.00g       0 
  /dev/sdc1  sdhp0_owl  lvm2 a--  508.00m       0 
  /dev/sdc2  timeshift  lvm2 a--   32.00g       0 
  /dev/sde1  sdhp0_owl  lvm2 a--  508.00m       0 
  /dev/sde2  timeshift  lvm2 a--   32.00g       0

There's one raid device (md2) and four LVM's (sdhp0_data,sdhp0,sdhp0_owl,timeshift). As you can see, a "production server" can be complex. For instance LVM "sdhp0_data" is an LVM sat upon /dev/md2 which is a raid device. Figuring it out with no docs from whoever configured the server can be hard.

[billtekker]

No worries. Yes, that's the raspberry pi operating system on the HP Proliant. The "sudo pvs" etc commands don't do anything but it claims that lvm2 is installed so I'm not sure how it's all configured. Maybe a partition was done and the raspberry pi is the boot drive? I'm in way above my head but need to open up the extra storage capacity for my business (time-lapse photography - Images are sent from cameras to our server).

Do you know anything about sym links? Wondering if I can create a sym link to the 2.4tb logical drive from the 240gb one without having to mount it maybe?

[trejan]

It isn't LVM. Your server has the HD attached to a HP SmartArray RAID controller. It is presumably an actual SmartArray card as they're SAS drives. The software SmartArray uses the onboard SATA controller and doesn't support SAS. You'll need to look at the HP documentation on how to manage the RAID controller. It is very much out of scope for this forum.

Using Raspberry Pi Desktop on a server is an unusual choice by your predecessor...

[swampdog]

@trjejan. Thanks for diving in. I knew something was amiss. Rpi desktop makes some sense.

@billtekker. I now know enough to know I don't know enough about your server to advise on anything, even if it was on-topic. All I can suggest is you confine yourself to non-harmful commands whilst on the HP server.

Code: Select all

$ cat /etc/fstab >/tmp/z
$ mount >/tmp/zz
$ lsblk >/tmp/zzz
$ df -hP >/tmp/zzzz

^^^print out those four files and try to find a suitable forum for that HP server.

I will now waffle on account I'm waiting for my PC to stop working..

Try and find the documentation on how to recover it. In days past when I used to be paid for administering black boxes in datacentres, we didn't backup a server in the sense us consumers think of. eg: if an oracle database server needed recovery the procedure we'd use would be we'd restore the OS, put the partitions in place, job done. The oracle DBA's would then use their docs to restore oracle. Officially we never knew how/what they did and vice-versa. The server is "irrelevant" in the sense that if the datacentre catches fire that database needs to be made running somewhere asap - who cares what server is being used.

I got my two ancient HP proliant servers cheap. There's software you can get which can administer them remotely or in this case (iirc) a web browser into a certain port. I never have known the credentials. Those are underneath. I set them up using a console then thought "bugger" when I found out. It has never been worth the trouble to take one down, extricate it and look underneath. I mention this because you will need such info. It can be a pita. We had some AIX servers (IBM unix) which would only talk to a limited version range(*) of firefox. Thus we had to maintain an air-gapped linux laptop which itself had to be backed up. I mention this on the off-chance you are tempted to fiddle: make sure you can talk to it beforehand.

(*) the one where if you typed "zuckerburg" into a text box then right-clicked for a spellcheck, it would suggest a crude word.

Now I don't need 24/7 uptime from my server room in the cellar I've slowly been decommissioning and migrating to rpi based solutions. The chip shortage has put that on hold but what I do have is one rpi4 with some usb enclosures attached. I've got quite a lot of 2tb and 3tb spinning rust sata disks from the HP servers. Basically if a disk out their raid failed, I'd replace all of the disks with bigger ones. My HP servers are not hot-swappable but there's a way (if you know mdadm) to remove the failed disk, shove in a "hot spare" and wait (many hours) for a rebuild. you then force-fail a good disk. Repeat until all disks replaced then "grow" the array.

The rpi4 enclosures spin down(**) a disk when not mounted. The plan was (still is) to just use plain ext4 on each disk in the enclosure and rsync one disk to another and rotate via a shelf. It's actually a bit more complex because I'm using LVM(***) on the rpi4 but same principle.

(**) except one disk which won't spin up. There's always an outlier. Same make/model as other disks. Slightly different firmware.

The key thing is how to recover this (currently) sole rpi4 on failure? Well I don't. It used to have an ssd attached. It doesn't need one just to be a "NAS". I've reverted to an sdcard which will fail and do so disastrously. I've shoved in a thumb drive and rsync the OS to that. I happen to use timeshift but other tools are available. Recovery will be as simple as writing whatever version of RaspOS onto a new sdcard, install timeshift, point it at the thumb drive and "restore".

(***) I used LVM to aggregate enough enclosure disks together to be large enough to rsync an entire HP server across (in case I screwed up). Each HP server was running (out of support) centos 5. Then I plugged an ssd into the esata port on the back of the HP server. Installed debian onto the ssd and had it boot off that. Installed samba,nfs,iscsi etc so it worked like before. Finally deleted the centos partition and added that space into LVM for data. Got a lot of space back because I had been using the OS partition as a local centos repo server. Repeat for other HP server. Nothing went wrong because I had a backup on the rpi4. Delete the rpi4 LVM. The downside is both are like the rpi4 - the OS has no raid. The upside, same scenario as the rpi4 on failure.

At work, when faced with a server with no docs, we'd decommission it. We'd virtualise it then take it off the network - see who squeals (timeframes of over a year). Thus, if you are the *only* one using this server, figure out how to reinstall it to your taste. Otherwise, wait a few weeks until rpi4 become available, buy one of those caddies you slot a couple of disks into, install (say) samba to point at one disk then create an rsync job to copy to the other disk. Periodically remove second disk (shelf), slot in third disk. Rotate.

Your only problem is finding the stable usb hardware. On my rpi4 I have the thumb drive plugged directly into one usb3 slot. The other usb3 slot is connected to one of those sabrent four port hubs with manual push buttons. I've got a few of those (two attached to my PC). The rpi4 enclosures are dangling off one of those, even though in that case I never use the manual buttons.

End of waffle. Hours later. Lots of interruptions. PC still working which is not a good thing. "GPU taken off the bus" in /var/log/syslog is what I wanted to catch but I'm watching it so of course it hasn't happened. C'est la vie!

[billtekker]

Ah right, thanks trejan, that makes sense. I tried signing up to the Hewlett Packard forum but it wouldn't allow me to and the customer service couldn't do it either, which is frustrating. I'll have to look for another forum.

Very helpful information about using raspberry pi's as backup drives there swampdog, thanks. I really want to be able to use the drives already installed but it's good to know anyway, for backup purposes (and maybe primary use if I can't get this sorted). Hope the PC behaves for you on your next attempt!

[billtekker]

It isn't LVM. Your server has the HD attached to a HP SmartArray RAID controller. It is presumably an actual SmartArray card as they're SAS drives. The software SmartArray uses the onboard SATA controller and doesn't support SAS. You'll need to look at the HP documentation on how to manage the RAID controller. It is very much out of scope for this forum.

I noticed when rebooting the server that the box for "smart array" isn't ticked.

SmartArray.jpg (248.73 KiB) Viewed 788 times

See red circle above.

Seems that it isn't using a smart array then, is that correct?

[trejan]

I noticed when rebooting the server that the box for "smart array" isn't ticked.

You must have a Smart Array if you've got SAS drives. It should be visible if you run "lspci".

[billtekker]

You're right, trejan. I noticed that when rebooting, the smart array box does actually get ticked right at the last second before switching to a different screen - I must have missed it before. Thanks for your advice.