• If you are still using CentOS 7.9, it's time to convert to Alma 8 with the free centos2alma tool by Plesk or Plesk Migrator. Please let us know your experiences or concerns in this thread:
    CentOS2Alma discussion
  • Please beaware of a breaking change in the REST API on the next Plesk release (18.0.62).
    Starting from Plesk Obsidian 18.0.62, requests to REST API containing the Content-Type header with a media-type directive other than “application/json” will result in the HTTP “415 Unsupported Media Type” client error response code. Read more here

Issue Disk speed slowed down from plesk?

MartinKth

New Pleskian
Server operating system version
Ubuntu 22.04
Plesk version and microupdate number
Plesk Obsidian Web Pro Edition Version 18.0.61 Update Nr. 5
Hi there,

I have a fresh server running Ubuntu 22.04 LTS (hetzner) - as we expect a lot traffic / database traffic, it's a pretty big machine. 192GB RAM, fast cpu, 6x NVMe disks in RAID 10.

When I tested the speed with
dd if=/dev/zero of=/root/testfile bs=50G count=5 oflag=direct
it showed me 1,7 GB/sec --> nice.
(I installed Alma before, the value was the same)
--> So, stable numbers, several tests, reboot - always around 1.7 GB/sec.


--> Now, after a clean Plesk install (via command line) - the speed is constantly down to 970 MB/sec --> a bit more than 50% from before. Nothing else has changed - rebooting etc. doesn't help. I've uninstalled most of the plesk extensions - but the number doesn't change.

RAID seems ok - /cat/mdstat shows:
root@.. ~ # cat /proc/mdstat
Personalities : [raid1] [raid10] [linear] [multipath] [raid0] [raid6] [raid5] [raid4]
md2 : active raid10 nvme1n1p4[0] nvme5n1p4[5] nvme0n1p4[1] nvme3n1p4[4] nvme4n1p4[3] nvme2n1p4[2]
5609204736 blocks super 1.2 512K chunks 2 near-copies [6/6] [UUUUUU]
bitmap: 10/42 pages [40KB], 65536KB chunk

md0 : active raid1 nvme1n1p2[0] nvme5n1p2[5] nvme0n1p2[1] nvme3n1p2[4] nvme4n1p2[3] nvme2n1p2[2]
4189184 blocks super 1.2 [6/6] [UUUUUU]

md1 : active raid1 nvme1n1p3[0] nvme5n1p3[5] nvme0n1p3[1] nvme3n1p3[4] nvme4n1p3[3] nvme2n1p3[2]
1046528 blocks super 1.2 [6/6] [UUUUUU]

unused devices: <none>


Does anybody have seen anything similar - is there a tip, how to solve this?
I explicitly chose Raid 10 instead RAID 5 or 6 to have a maximum writing speed...

Thanks - BR
Martin
 
A server under no load can reserve a maximum of cpu and io power to the disk transactions. A server with software on it will need to do a few other things, too, hence the same test will result in less throughput. You can use the Linux iotop command to see whether the server is busy with other disk read/write tasks.
 
Thanks for the tip.
But iotop shows - the server is just idling. iotop toggles between a few K/sec and 0.00 K/sec.

I don't have any domains/customers/databases in plesk yes. It's just a clean install.
For me it looks like plesk changed the disk configuration somehow. But I have no idea where to look closer. fstab etc. looks normal.

Does anyone have a tip for me?
 

Attachments

  • Screenshot 2024-06-14 at 08.46.29.jpg
    Screenshot 2024-06-14 at 08.46.29.jpg
    149.7 KB · Views: 0
What change to a disk configuration could cause a slower disk? Even considering that a software changed something (which Plesk does not do) I would not know how to tell a drive to work slower than it can.
 
Since its a clean server, with nothing running in Plesk, perhaps try uninstalling Plesk to see if your performance returns to the pre-Plesk numbers. Then re-install Plesk and check the numbers again. Not saying it is or is not Plesk, but that's what I would do.
 
Back
Top