[linux-lvm] Cache block size warning

[Ciprian Hacman]

Hi,

We are configuring lvm with SSD caching on some of our boxes in AWS. When
the cache is initialised, we see this in logs:

device-mapper*: cache: You have created a cache device with a lot of
individual cache blocks (4915200)*

* All these mappings can consume a lot of kernel memory, and
take some time to read/write.*

* Please consider increasing the cache block size to reduce
the overall cache block count.*


Is this something to worry? If so, how can we improve it?

For reference, the output of lsblk is:

NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT

xvda 202:0 0 25G 0 disk

└─xvda1 202:1 0 25G 0 part /

xvdb 202:16 0 152.6G 0 disk

└─md0 9:0 0 304.9G 0 raid0

├─storage-cache_cdata 252:0 0 300G 0 lvm

│ └─storage-data 252:3 0 1.5T 0 lvm /mnt/local

└─storage-cache_cmeta 252:1 0 1G 0 lvm

└─storage-data 252:3 0 1.5T 0 lvm /mnt/local

xvdc 202:32 0 152.6G 0 disk

└─md0 9:0 0 304.9G 0 raid0

├─storage-cache_cdata 252:0 0 300G 0 lvm

│ └─storage-data 252:3 0 1.5T 0 lvm /mnt/local

└─storage-cache_cmeta 252:1 0 1G 0 lvm

└─storage-data 252:3 0 1.5T 0 lvm /mnt/local

xvdg 202:96 0 1.5T 0 disk

└─storage-data_corig 252:2 0 1.5T 0 lvm

└─storage-data 252:3 0 1.5T 0 lvm /mnt/local


Thanks,
Ciprian
--
Performance Monitoring * Log Analytics * Search Analytics
Solr & Elasticsearch Support * http://sematext.com/

[Zdenek Kabelac]

Hi,
We are configuring lvm with SSD caching on some of our boxes in AWS. When the
device-mapper*: cache: You have created a cache device with a lot of
individual cache blocks (4915200)*
* All these mappings can consume a lot of kernel memory, and
take some time to read/write.*
* Please consider increasing the cache block size to reduce
the overall cache block count.*
Is this something to worry? If so, how can we improve it?

If you have lots of kernel memory - it's probably ok.
Otherwise you may try to use bigger chunk size.

e.g. --chunksize 512K or even more:

lvcreate --type cache-pool -L300G vg/cpool

lvcreate -H --chunksize 512K -L1.5T --name data vg/cpool

It's possible future version of lvm2 will try to create this by default.

Zdenek

[Ciprian Hacman]

Thank you for the quick answer. Sounds easy to implement if needed.

Ciprian
--
Performance Monitoring * Log Analytics * Search Analytics
Solr & Elasticsearch Support * http://sematext.com/

Post by Zdenek Kabelac

Hi,
We are configuring lvm with SSD caching on some of our boxes in AWS. When the
device-mapper*: cache: You have created a cache device with a lot of
individual cache blocks (4915200)*
* All these mappings can consume a lot of kernel memory, and
take some time to read/write.*
* Please consider increasing the cache block size to reduce
the overall cache block count.*
Is this something to worry? If so, how can we improve it?

If you have lots of kernel memory - it's probably ok.
Otherwise you may try to use bigger chunk size.
lvcreate --type cache-pool -L300G vg/cpool
lvcreate -H --chunksize 512K -L1.5T --name data vg/cpool
It's possible future version of lvm2 will try to create this by default.
Zdenek
_______________________________________________
linux-lvm mailing list
https://www.redhat.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://tldp.org/HOWTO/LVM-HOWTO/