crushmap使用说明

This article introduces a simple use case for storage providers. For some reasons some customers would like to pay more for a fast storage solution and other would prefer to pay less for a reasonnable storage solution.

I. Use case

Roughly say your infrastructure could be based on several type of servers:

• storage nodes full of SSDs disks

• storage nodes full of SAS disks

• storage nodes full of SATA disks

Such handy mecanism is possible with the help of the CRUSH Map.

II. A bit about CRUSH

CRUSH stands for Controlled Replication Under Scalable Hashing:

• Pseudo-random placement algorithm

• Fast calculation, no lookup Repeatable, deterministic

• Ensures even distribution

• Stable mapping

• Limited data migration

• Rule-based configuration, rule determines data placement

• Infrastructure topology aware, the map knows the structure of your infra (nodes, racks, row, datacenter)

• Allows weighting, every OSD has a weight

For more details check the Ceph Official documentation.

III. Setup

What are we going to do?

1. Retrieve the current CRUSH Map

2. Decompile the CRUSH Map

3. Edit it. We will add 2 buckets and 2 rulesets

4. Recompile the new CRUSH Map.

5. Re-inject the new CRUSH Map.

III.1. Begin

Grab your current CRUSH map:

12	$ ceph osd getcrushmap -o ma-crush-map$ crushtool -d ma-crush-map -o ma-crush-map.txt

For the sake of simplicity, let’s assume that you have 4 OSDs:

• 2 of them are SAS disks

• 2 of them are SSD enterprise

And here is the OSD tree:

1234567891011

$ ceph osd treedumped osdmap tree epoch 621# id    weight  type name   up/down reweight-1  12  pool default-3  12      rack le-rack-2  3           host ceph-010   1               osd.0   up  11   1               osd.1   up  1-4  3           host ceph-022   1               osd.2   up  13   1               osd.3   up  1

III.2. Default crush map

Edit your CRUSH map:

# begin crush map

# devices
device 0 osd.0
device 1 osd.1
device 2 osd.2
device 3 osd.3

# types
type 0 osd
type 1 host
type 2 rack
type 3 row
type 4 room
type 5 datacenter
type 6 pool

# buckets
host ceph-01 {
    id -2       # do not change unnecessarily
    # weight 3.000
    alg straw
    hash 0  # rjenkins1
    item osd.0 weight 1.000
    item osd.1 weight 1.000
}
host ceph-02 {
    id -4       # do not change unnecessarily
    # weight 3.000
    alg straw
    hash 0  # rjenkins1
    item osd.2 weight 1.000
    item osd.3 weight 1.000
}
rack le-rack {
    id -3       # do not change unnecessarily
    # weight 12.000
    alg straw
    hash 0  # rjenkins1
    item ceph-01 weight 2.000
    item ceph-02 weight 2.000
}
pool default {
    id -1       # do not change unnecessarily
    # weight 12.000
    alg straw
    hash 0  # rjenkins1
    item le-rack weight 4.000
}

# rules
rule data {
    ruleset 0
    type replicated
    min_size 1
    max_size 10
    step take default
    step chooseleaf firstn 0 type host
    step emit
}
rule metadata {
    ruleset 1
    type replicated
    min_size 1
    max_size 10
    step take default
    step chooseleaf firstn 0 type host
    step emit
}
rule rbd {
    ruleset 2
    type replicated
    min_size 1
    max_size 10
    step take default
    step chooseleaf firstn 0 type host
    step emit
}

# end crush map

III.3. Add buckets and rules

Now we have to add 2 new specific rules:

• one for the SSD pool

• one for the SAS pool

III.3.1. SSD Pool

Add a bucket for the pool SSD:

pool ssd {
    id -5       # do not change unnecessarily
    alg straw
    hash 0  # rjenkins1
    item osd.0 weight 1.000
    item osd.1 weight 1.000
}

Add a rule for the bucket nearly created:

rule ssd {
    ruleset 3
    type replicated
    min_size 1
    max_size 10
    step take ssd
    step choose firstn 0 type osd
    step emit
}

III.3.1. SAS Pool

Add a bucket for the pool SAS:

pool sas {
    id -6       # do not change unnecessarily
    alg straw
    hash 0  # rjenkins1
    item osd.2 weight 1.000
    item osd.3 weight 1.000
}

Add a rule for the bucket nearly created:

rule sas {
    ruleset 4
    type replicated
    min_size 1
    max_size 10
    step take sas
    step choose firstn 0 type osd
    step emit
}

Eventually recompile and inject the new CRUSH map:

12	$ crushtool -c ma-crush-map.txt -o ma-nouvelle-crush-map$ ceph osd setcrushmap -i ma-nouvelle-crush-map

III.3. Create and configure the pools

Create your 2 new pools:

1234	$ rados mkpool ssdsuccessfully created pool ssd$ rados mkpool sassuccessfully created pool sas

Set the rule set to the pool:

12	ceph osd pool set ssd crush_ruleset 3ceph osd pool set sas crush_ruleset 4

Check that the changes have been applied successfully:

123	$ ceph osd dump | grep -E 'ssd|sas'pool 3 'ssd' rep size 2 crush_ruleset 3 object_hash rjenkins pg_num 128 pgp_num 128 last_change 21 owner 0pool 4 'sas' rep size 2 crush_ruleset 4 object_hash rjenkins pg_num 128 pgp_num 128 last_change 23 owner 0

III.4. Test it

Just create some random files and put them into your object store:

1234	$ dd if=/dev/zero of=ssd.pool bs=1M count=512 conv=fsync$ dd if=/dev/zero of=sas.pool bs=1M count=512 conv=fsync$ rados -p ssd put ssd.pool ssd.pool.object$ rados -p sas put sas.pool sas.pool.object

Where are pg active?

12345

$ ceph osd map ssd ssd.pool.objectosdmap e260 pool 'ssd' (3) object 'ssd.pool.object' -> pg 3.c5034eb8 (3.0) -> up [1,0] acting [1,0]$ ceph osd map sas sas.pool.objectosdmap e260 pool 'sas' (4) object 'sas.pool.object' -> pg 4.9202e7ee (4.0) -> up [3,2] acting [3,2]

CRUSH Rules! As you can see from this article CRUSH allows you to perform amazing things. The CRUSH Map could be very complex, but it brings a lot of flexibility! Happy CRUSH Mapping ;-)