Chapter 3. Placement Groups (PGs)

Ceph strives to prevent the permanent loss of data. However, after an OSD fails, the risk of permanent data loss increases until the data it had is fully recovered. Permanent data loss, though rare, is still possible. The following scenario describes how Ceph could permanently lose data in a single placement group with three copies of the data:

Hardware failure isn’t an exception, but an expectation. To prevent the foregoing scenario, ideally the recovery process should be as fast as reasonably possible. The size of your cluster, your hardware configuration and the number of placement groups play an important role in total recovery time.

It is likely that Ceph stored the remaining 150 placement groups randomly across the 9 remaining OSDs. Therefore, each remaining OSD is likely to send copies of objects to all other OSDs and also receive some new objects, because the remaining OSDs become responsible for some of the 150 placement groups now assigned to them.

The total recovery time depends upon the hardware supporting the pool. For example, in a 10 OSD cluster, if a host contains one OSD with a 1 TB SSD, and a 10 GB/s switch connects each of the 10 hosts, the recovery time will take M minutes. By contrast, if a host contains two SATA OSDs and a 1 GB/s switch connects the five hosts, recovery will take substantially longer. Interestingly, in a cluster of this size, the number of placement groups has almost no influence on data durability. The placement group count could be 128 or 8192 and the recovery would not be slower or faster.

However, growing the same Ceph cluster to 20 OSDs instead of 10 OSDs is likely to speed up recovery and therefore improve data durability significantly. Why? Each OSD now participates in only 75 placement groups instead of 150. The 20 OSD cluster will still require all 19 remaining OSDs to perform the same amount of copy operations in order to recover. In the 10 OSD cluster, each OSDs had to copy approximately 100 GB. In the 20 OSD cluster each OSD only has to copy 50 GB each. If the network was the bottleneck, recovery will happen twice as fast. In other words, recovery time decreases as the number of OSDs increases.

In the 10 OSD cluster described above, if any OSD fails, then approximately 8 placement groups (that is 75 pgs / 9 osds being recovered) will only have one surviving copy. And if any of the 8 remaining OSDs fail, the last objects of one placement group are likely to be lost (that is 8 pgs / 8 osds with only one remaining copy being recovered). This is why starting with a somewhat larger cluster is preferred (for example, 50 OSDs).

When the size of the cluster grows to 20 OSDs, the number of placement groups damaged by the loss of three OSDs drops. The second OSD lost will degrade approximately 2 (that is 35 pgs / 19 osds being recovered) instead of 8 and the third OSD lost will only lose data if it is one of the two OSDs containing the surviving copy. In other words, if the probability of losing one OSD is 0.0001% during the recovery time frame, it goes from 8 * 0.0001% in the cluster with 10 OSDs to 2 * 0.0001% in the cluster with 20 OSDs. Having 512 or 4096 placement groups is roughly equivalent in a cluster with less than 50 OSDs as far as data durability is concerned.

When you add an OSD to the cluster, it might take a long time to populate the new OSD with placement groups and objects. However there is no degradation of any object and adding the OSD has no impact on data durability.

Ceph seeks to avoid hot spots—​that is, some OSDs receive substantially more traffic than other OSDs. Ideally, CRUSH assigns objects to placement groups evenly so that when the placement groups get assigned to OSDs (also pseudo randomly), the primary OSDs store objects such that they are evenly distributed across the cluster and hot spots and network over-subscription problems cannot develop because of data distribution.

Since CRUSH computes the placement group for each object, but does not actually know how much data is stored in each OSD within this placement group, the ratio between the number of placement groups and the number of OSDs might influence the distribution of the data significantly.

For instance, if there was only one placement group with ten OSDs in a three replica pool, Ceph would only use three OSDs to store data because CRUSH would have no other choice. When more placement groups are available, CRUSH is more likely to evenly spread objects across OSDs. CRUSH also evenly assigns placement groups to OSDs.

As long as there are one or two orders of magnitude more placement groups than OSDs, the distribution should be even. For instance, 256 placement groups for 3 OSDs, 512 or 1024 placement groups for 10 OSDs, and so forth.

The ratio between OSDs and placement groups usually solves the problem of uneven data distribution for Ceph clients that implement advanced features like object striping. For example, a 4 TB block device might get sharded up into 4 MB objects.

The ratio between OSDs and placement groups does not address uneven data distribution in other cases, because CRUSH does not take object size into account. Using the librados interface to store some relatively small objects and some very large objects can lead to uneven data distribution. For example, one million 4K objects totaling 4 GB are evenly spread among 1000 placement groups on 10 OSDs. They will use 4 GB / 10 = 400 MB on each OSD. If one 400 MB object is added to the pool, the three OSDs supporting the placement group in which the object has been placed will be filled with 400 MB + 400 MB = 800 MB while the seven others will remain occupied with only 400 MB.

For each placement group, OSDs and Ceph monitors need memory, network and CPU at all times, and even more during recovery. Sharing this overhead by clustering objects within a placement group is one of the main reasons placement groups exist.

Minimizing the number of placement groups saves significant amounts of resources.

The PG calculator calculates the number of placement groups for you and addresses specific use cases. The PG calculator is especially helpful when using Ceph clients like the Ceph Object Gateway where there are many pools typically using the same rule (CRUSH hierarchy). You might still calculate PGs manually using the guidelines in PG Count for Small Clusters and Calculating PG Count. However, the PG calculator is the preferred method of calculating PGs.

See Ceph Placement Groups (PGs) per Pool Calculator on the Red Hat Customer Portal for details.

When you create a pool, you also create a number of placement groups for the pool. If you don’t specify the number of placement groups, Ceph will use the default value of 8, which is unacceptably low. You can increase the number of placement groups for a pool, but we recommend setting reasonable default values in your Ceph configuration file too.

osd pool default pg num = 100
osd pool default pgp num = 100

You need to set both the number of placement groups (total), and the number of placement groups used for objects (used in PG splitting). They should be equal.

Small clusters don’t benefit from large numbers of placement groups. As the number of OSDs increase, choosing the right value for pg_num and pgp_num becomes more important because it has a significant influence on the behavior of the cluster as well as the durability of the data when something goes wrong (that is the probability that a catastrophic event leads to data loss). It is important to use the PG calculator with small clusters.

If you have more than 50 OSDs, we recommend approximately 50-100 placement groups per OSD to balance out resource usage, data durability and distribution. If you have less than 50 OSDs, choosing among the PG Count for Small Clusters is ideal. For a single pool of objects, you can use the following formula to get a baseline:

                (OSDs * 100)
   Total PGs =  ------------
                 pool size

Where pool size is either the number of replicas for replicated pools or the K+M sum for erasure coded pools (as returned by ceph osd erasure-code-profile get).

You should then check if the result makes sense with the way you designed your Ceph cluster to maximize data durability, data distribution and minimize resource usage.

The result should be rounded up to the nearest power of two. Rounding up is optional, but recommended for CRUSH to evenly balance the number of objects among placement groups.

For a cluster with 200 OSDs and a pool size of 3 replicas, you would estimate your number of PGs as follows:

   (200 * 100)
   ----------- = 6667. Nearest power of 2: 8192
        3

With 8192 placement groups distributed across 200 OSDs, that evaluates to approximately 41 placement groups per OSD. You also need to consider the number of pools you are likely to use in your cluster, since each pool will create placement groups too. Ensure that you have a reasonable maximum PG count.

When using multiple data pools for storing objects, you need to ensure that you balance the number of placement groups per pool with the number of placement groups per OSD so that you arrive at a reasonable total number of placement groups. The aim is to achieve reasonably low variance per OSD without taxing system resources or making the peering process too slow.

In an exemplary Ceph Storage Cluster consisting of 10 pools, each pool with 512 placement groups on ten OSDs, there are a total of 5,120 placement groups spread over ten OSDs, or 512 placement groups per OSD. That might not use too many resources depending on your hardware configuration. By contrast, if you create 1,000 pools with 512 placement groups each, the OSDs will handle ~50,000 placement groups each and it would require significantly more resources. Operating with too many placement groups per OSD can significantly reduce performance, especially during rebalancing or recovery.

The Ceph Storage Cluster has a default maximum value of 300 placement groups per OSD. You can set a different maximum value in your Ceph configuration file.

mon pg warn max per osd

Each pool in the Red Hat Ceph Storage cluster has a pg_autoscale_mode property for PGs that you can set to off, on, or warn.

The autoscaler uses the bulk flag to determine which pool should start with a full complement of PGs and only scales down when the usage ratio across the pool is not even. However, if the pool does not have the bulk flag, the pool starts with minimal PGs and only when there is more usage in the pool.

You can view the pool, it’s relative utilization and any suggested changes to the PG count in the storage cluster.

SIZE is the amount of data stored in the pool.

TARGET SIZE, if present, is the amount of data the administrator has specified they expect to eventually be stored in this pool. The system uses the larger of the two values for its calculation.

RATE is the multiplier for the pool that determines how much raw storage capacity the pool uses. For example, a 3 replica pool has a ratio of 3.0, while a k=4,m=2 erasure coded pool has a ratio of 1.5.

RAW CAPACITY is the total amount of raw storage capacity on the OSDs that are responsible for storing the pool’s data.

RATIO is the ratio of the total capacity that the pool is consuming, that is, ratio = size * rate / raw capacity.

TARGET RATIO, if present, is the ratio of storage the administrator has specified that they expect the pool to consume relative to other pools with target ratios set. If both target size bytes and ratio are specified, the ratio takes precedence. The default value of TARGET RATIO is 0 unless it was specified while creating the pool. The more the --target_ratio you give in a pool, the larger the PGs you are expecting the pool to have.

EFFECTIVE RATIO, is the target ratio after adjusting in two ways: 1. subtracting any capacity expected to be used by pools with target size set. 2. normalizing the target ratios among pools with target ratio set so they collectively target the rest of the space. For example, 4 pools with target ratio 1.0 would have an effective ratio of 0.25. The system uses the larger of the actual ratio and the effective ratio for its calculation.

BIAS, is used as a multiplier to manually adjust a pool’s PG based on prior information about how much PGs a specific pool is expected to have. By default, the value if 1.0 unless it was specified when creating a pool. The more --bias you give in a pool, the larger the PGs you are expecting the pool to have.

PG_NUM is the current number of PGs for the pool, or the current number of PGs that the pool is working towards, if a pg_num change is in progress. NEW PG_NUM, if present, is the suggested number of PGs (pg_num). It is always a power of 2, and is only present if the suggested value varies from the current value by more than a factor of 3.

AUTOSCALE, is the pool pg_autoscale_mode, and is either on, off, or warn.

BULK, is used to determine which pool should start out with a full complement of PGs. BULK only scales down when the usage ratio cross the pool is not even. If the pool does not have this flag the pool starts out with a minimal amount of PGs and only used when there is more usage in the pool.

The BULK values are true, false, 1, or 0, where 1 is equivalent to true and 0 is equivalent to false. The default value is false.

Set the BULK value either during or after pool creation.

For more information about using the bulk flag, see Create a pool and Setting placement group auto-scaling modes.

Allowing the cluster to automatically scale PGs based on cluster usage is the simplest approach to scaling PGs. Red Hat Ceph Storage takes the total available storage and the target number of PGs for the whole system, compares how much data is stored in each pool, and apportions the PGs accordingly. The command only makes changes to a pool whose current number of PGs (pg_num) is more than three times off from the calculated or suggested PG number.

The target number of PGs per OSD is based on the mon_target_pg_per_osd configurable. The default value is set to 100.

ceph osd pool create --pg-num-min NUMBER
ceph osd pool create --pg-num-max NUMBER

ceph osd pool create --pg-num-min 50
ceph osd pool create --pg-num-max 150

If you want to enable or disable the autoscaler for all the pools at the same time, you can use the noautoscale global flag. This global flag is useful during upgradation of the storage cluster when some OSDs are bounced or when the cluster is under maintenance. You can set the flag before any activity and unset it once the activity is complete.

By default, the noautoscale flag is set to off. When this flag is set, then all the pools have pg_autoscale_mode as off and all the pools have the autoscaler disabled.

A newly created pool consumes a small fraction of the total cluster capacity and appears to the system that it will need a small number of PGs. However, in most cases, cluster administrators know which pools are expected to consume most of the system capacity over time. If you provide this information, known as the target size to Red Hat Ceph Storage, such pools can use a more appropriate number of PGs (pg_num) from the beginning. This approach prevents subsequent changes in pg_num and the overhead associated with moving data around when making those adjustments.

You can specify target size of a pool in these ways:

To set the number of placement groups in a pool, you must specify the number of placement groups at the time you create the pool. See Create a Pool for details. Once you set placement groups for a pool, you can increase or decrease the number of placement groups. To change the number of placement groups, use the following command:

ceph osd pool set POOL_NAME pg_num PG_NUMBER

[ceph: root@host01 /]# ceph osd pool set pool1 pg_num 60
set pool 2 pg_num to 60

Once you increase or decrease the number of placement groups, you must also adjust the number of placement groups for placement (pgp_num) before your cluster will rebalance. The pgp_num should be equal to the pg_num.

ceph osd pool set POOL_NAME pgp_num PGP_NUMBER

[ceph: root@host01 /]# ceph osd pool set pool1 pgp_num 60
set pool 2 pgp_num to 60

Get the number of placement groups in a pool:

ceph osd pool get POOL_NAME pg_num

[ceph: root@host01 /]# ceph osd pool get pool1 pg_num
pg_num: 60

Get statistics for the placement groups:

ceph pg dump [--format FORMAT]

Valid formats are plain (default) and json.

Get the statistics for all placement groups stuck in a specified state:

ceph pg dump_stuck {inactive|unclean|stale|undersized|degraded [inactive|unclean|stale|undersized|degraded...]} INTEGER

Inactive Placement groups cannot process reads or writes because they are waiting for an OSD with the most up-to-date data to come up and in.

Unclean Placement groups contain objects that are not replicated the desired number of times. They should be recovering.

Stale Placement groups are in an unknown state - the OSDs that host them have not reported to the monitor cluster in a while (configured by mon_osd_report_timeout).

Valid formats are plain (default) and json. The threshold defines the minimum number of seconds the placement group is stuck before including it in the returned statistics (default 300 seconds).

Get the placement group map for a particular placement group:

ceph pg map PG_ID

[ceph: root@host01 /]# ceph pg map 1.6c
osdmap e13 pg 1.6c (1.6c) -> up [1,0] acting [1,0]

Ceph returns the placement group map, the placement group, and the OSD status.

Retrieve statistics for a particular placement group:

ceph pg PG_ID query

Scrub a placement group:

ceph pg scrub PG_ID

Ceph checks the primary and any replica nodes, generates a catalog of all objects in the placement group, and compares them to ensure that no objects are missing or mismatched, and their contents are consistent. Assuming the replicas all match, a final semantic sweep ensures that all of the snapshot-related object metadata is consistent. Errors are reported via logs.

If the cluster has lost one or more objects, and you have decided to abandon the search for the lost data, you must mark the unfound objects as lost.

If all possible locations have been queried and objects are still lost, you might have to give up on the lost objects. This is possible given unusual combinations of failures that allow the cluster to learn about writes that were performed before the writes themselves are recovered.

Ceph supports only "revert" option, which either rolls back to a pervious version of the object or forgets about it if it was a new object. To mark the "unfound" objects as "lost", use the following command:

ceph pg PG_ID mark_unfound_lost revert|delete