Good morning,
I have a DL380, I would like to put the disk where the storj data are stored in RAID 1, is it possible to do this when the disk where the data is is already declared in the HP disk manager?
Thanks
About what amount of data are we talking here?
raid1 makes no sense, because it halves earnings / doubles expenses.
3 Likes
Sorry, I misspoke, I want to put the disk where Proxmox is stored in RAID1, so that in the event of a failure of the disk storing Proxmox, I can switch to the other
@daki82 40GB
so a fresh node?
knowing nothing about DL380 (maybe ask at DL380 forum)i would:
stop the node,
move data to spare drive/usbstick
initialize raid1
move data back
even if you lose the node, its propably 3 days old?
or are we talking about the os disk?
mirrors / raid1 is basically just a copy of the disk… so you should be able to do that with no downtime or anything…
consult the manual of whatever you are using for your storage…
hardware raid, software raid… i would prefer software, but i suspect that in an old HP system you will most likely be using a raid card.
i would suspect an LSI one, you would access the megaraid software and just add a mirror to your OS disk.
…
but yeah check your manual or similar guides.
should be very straight forward.
mirror are always a good idea for OS drives.
1 Like
That I something I strongly disagree with. OS drives don’t contain anything important nor performance sensitive. Using redundancy on a os/boot drive is a massive overkill. Especially the most wasteful type of redundancy.
Yes it’s almost universally possible without downtime with pretty much any appliances I have encountered so far.
But not advisable, if you are only doing it for storage node. Instead, start a second node on the second drive.
If you are doing it for other reasons — then sure.
2 Likes
If you have significant customisations or custom scripts that run it can be cheaper to have the redundancy than needing the time to rebuild due to drive failure.
Of course redundancy doesn’t stop failure - note the incident that happened to “serve the home” 9-Step Calm and Easy Proxmox VE Boot Drive Failure Recovery
I usually uses git or restic to store such customizations - very easy to restore. Usually one hour is enough to install the OS and apply customizations.
For Windows you may use their integrated backup software, it allows to partially restore, or in edge case you always can mount a vhdx disk from it.
1 Like
If I only use restic, what should I save from Proxmox? Which configuration file etc…?
Disk failure is just one of many ways your system can fail. You are essentially paying cost of another drive, opportunity cost of an extra data port, plus electricity, to protect from this single mode of failure.
All while there is another way to protect against all failure modes that does not cost anything.
Setup the system, snapshot it, and if it fails regardless of the reason — restore the snapshot to the new drive.
Some systems (like TrueNAS) take this on a next level where the system state including all customizations are recorded to a script that can be played back on the vanilla OS to bring it to that state. In this case you just export that script. My TrueNAS system boots from a $2 thumb drive from Amazon for the past two years. Not a raid one 
Yes, but what do I currently save from the Proxmox configuration?
But perhaps it’s better to use an integrated method: Backup and Restore - Proxmox VE
You may use Storj DCS to store backups.
So, having come from a situation where my employer booted his esxi servers from thumb drives - that is not a situation I want to go back to. That said, I boot my TrueNAS boxes from a single SSD - not a redundant one.
Of course it is, but that is why servers come with things like dual power supplies. Not everything is redundant of course.
The redundancy is more to give you time to respond to failure. For example our Datacenter servers are quite a few kilometers away from the nearest staff member. To get someone on site to replace a failed part takes time. Redundant parts often has given us that time to organise to get someone in the DC to do the required work without an unplanned service outage… Without it we would just be down - until someone can get to the DC.