Project

General

Profile

Actions

Task #5770

closed

Create a howto: how we maintain our disks and ceph cluster(s)

Added by Nico Schottelius about 6 years ago. Updated over 5 years ago.

Status:
Rejected
Priority:
Normal
Assignee:
Target version:
-
Start date:
10/05/2018
Due date:
10/08/2018
% Done:

0%

Estimated time:
PM Check date:

Description

  • Detailed explanation
  • Including commands of megacli
  • Including motivation

Disk handling

Our servers are using Perc H700 or H800 disk controllers. The tool megacli can be used to manage them.

Handling foreign configuration

When plugging a disk into a server that was configured / used in a different server before, we need to clear the previous configuration.

Listing all known foreign configuration using megacli -CfgForeign -Scan -aALL.

Remove/clear the foreign configuration using

# for instance -L5 -a0 or -L2 -a1
megacli -DiscardPreservedCache -L<slot-or-ID> -a<Controller>

Removal of a disk

If a (dead or intact) disk is being removed from the system, the system will HANG on the next reboot, complaining about it missing. For that reason AFTER removing a disk, we need to "clear the cache":


# Get list of caches that need to be cleared
megacli -GetPreservedCacheList -aALL

# Clear cache: -L0 -a1
megacli -DiscardPreservedCache -L<??> -a<controller>

How to move an OSD (Disk, SSD) to another server

  • Use ceph-osd-stop-disable from ungleich tools to stop / disable the OSD
  • Clear the raid controller information
  • Insert disk into new host
  • Clear the foreign configuration (megacli)
  • [add it to the system with megacli]
  • OSD should automatically come up afterwards
    • Check using ceph osd tree
Actions

Also available in: Atom PDF