All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Anyone with RAID-1/ax101 willing to help me ?
Background
Long story short, I create this bench script written in shell that works similarly to yabs— except I employ additional feature and one of them being able to support multi disk layout.
So a user just reported issues with the script, he said there is a formatting problem. I have look into the code and first thing I thought was problem with loop control (this is the code if anyone want to take a look → https://github.com/L1so/benchy/blob/ec2649009f0e878304ef6f8b73e6ef843fe504ef/benchy#L477).
Later I realized I was wrong and actually the original code is correct, continue instead of continue 2 because the function is placed on outer loop and not nested one.
Actual problem
I decided to look back on the issues and found the culprit, it lies with RAID-1 configuration that the user was using in his hetzner ax101 dedi, it create a mirror of a disk to another disk so there was two disk hold the same data. My suspicion is confirmed when I ask the user to run a command to troubleshoot the problem.
Before I PM him, he also provide lsblk and df -h output, at that time I haven't noticed raid1 type from his lsblk.
With that being said, if anyone has access to a server with raid1 or ax101 itself, I kindly ask you to help me diagnose this problem. You can participate by running this command and post the output.
lsblk -nPpbo KNAME,SIZE,PKNAME,MOUNTPOINT,FSTYPE
With that, I will try eliminate the duplicate raid 1 disk with awk.
Yeah I can just pipe the output to this, but that feels like the wrong way and I probably will use that as last resort.
Thanks !
Comments
I think this hasn't anything to do with the server type (or provider). It's simply how md/soft-raid works.
lsblk doesn't help much in that regard as it lists all block devices and that kind of raid sits on top of it.
here is an lsblk from a system using a mix of soft-raid (raid10) and zfs:
I'd suspect the depending on the zpool config you can also run into troubles.
may I ask, why you don't parse
mountoutput for usable/benchable dirs? ;-)I'm trying to fetch following data: partition name, mountpoint, size, filesystem type.
lsblkprovide that four data whilemountonly has 3 of them, with no size information in the output (though I can easily feed the mountpoint todfto fetch its size).If it's okay with you, can you show me
mountoutput in your system ?sure, same box:
PS: and yes, I understand the thing with the sizes in lsblk being convenient, however with things like raid-10 or even raid-5 raid-6 this would not work, as you'll only see the size of each block-device and not the mounted size. apart from that there possibly is also LVM or btrfs which allow for their own 'grouping' ...
what about parsing
df -T? otherwise yes, you'll need at least two requests for those four fields ;-)