19 posts tagged with "k8s"

For a client, we are developing a unified interface/service to manage Virtual Machines and Containers for the Private Cloud with Kubernetes, Openstack, CloudStack, Opennebula, VMWare vSphere and Hyperscalers, such as GCP, AWS and Azure.

We are currently focusing on the following advantages:

• Centralised management of distributed resources
• Evaluation of costs across different providers
• Reconciliation and implementation of planning
• Mitigation of cloud provider failures
• Connection of egde computing
• Connection of sites

Since 2021, we have supported one of our customers in the architecture and operation of more than 150 Kubernetes clusters with Ceph worldwide in all hyperscalers, such as GCP, AWS, Microsoft Azure and AliCloud.

Use of programming languages to define customised infrastructure environments using Terraform. Use of language constructs and tools that follow Infrastructure as Code (IoC) design patterns.

Use of functions and libraries within programming languages to develop complex infrastructure projects.

Use programming languages such as TypeScript, Python and Go to map multi-cloud environments. With the modular and open architecture to include hundreds of vendors and thousands of module definitions.

After more than 10 years of experience with different evolutionary stages of timeseries monitoring systems and the growing need for metrics, we decided to replace database and file based solutions.

To store metrics for the long term, we rely on object storage technology to provide almost unlimited storage capacity in the backend.

The Object Storage is provided to us by multiple Ceph clusters. We are now also able to dynamically connect alternative storage locations, such as AWS and/or GCP, as needed.

validate if the RBD Cache is active on your client​

By default the cache is enabled, since Version 0.87.

To enable the cache on the client side you have to add following config /etc/ceph/ceph.conf:

[client]
rbd cache = true
rbd cache writethrough until flush = true

add local admin socket​

So that you can also verify the status on the client side, you must add the following two parameters:

[client]
admin socket = /var/run/ceph/$cluster-$type.$id.$pid.$cctid.asok
log file = /var/log/ceph/

configure permissions and security​

Both paths must be writable by the user who uses the RBD library. Applications such as SELinux or AppArmor must be properly configured.

request infos via admin socket​

Once this is done, run your application that is supposed to use librbd (kvm, docker, podman, ...) and request the information via the admin daemon socket:

$ sudo ceph --admin-daemon /var/run/ceph/ceph-client.admin.66606.140190886662256.asok config show | grep rbd_cache "rbd_cache": "true", "rbd_cache_writethrough_until_flush": "true", "rbd_cache_size": "33554432", "rbd_cache_max_dirty": "25165824", "rbd_cache_target_dirty": "16777216", "rbd_cache_max_dirty_age": "1", "rbd_cache_max_dirty_object": "0", "rbd_cache_block_writes_upfront": "false",

Verify the cache behaviour​

To compare the performance difference you can test the cache, you can deactivate it in the [client] section in your ceph.conf as follows:

[client]
rbd cache = false

Then run a fio benchmark with the following command:

fio --ioengine=rbd --pool=<pool-name> --rbdname=rbd1 --direct=1 --fsync=1 --rw=write --bs=4k --numjobs=1 --iodepth=1 --runtime=60 --time_base

Finally, run this test with RBD client cache enabled and disabled and you should notice a significant difference.

Sources​

https://www.sebastien-han.fr/blog/2015/09/02/ceph-validate-that-the-rbd-cache-is-active/

We were speakers at the first edition of Cloudland

Cloudland is the festival of the German-speaking Cloud Native Community (DCNC), with the aim of communicating the current status quo in the use of cloud technologies and focusing in particular on future challenges.

Our contribution on Multi Cloud Deployment met with great interest at the "Container & Cloud Technologies" theme day.

What are hugepages?​

For example, x86 CPUs normally support 4K and 2M (1G if architecturally supported) page sizes, ia64 architecture supports multiple page sizes 4K, 8K, 64K, 256K, 1M, 4M, 16M, 256M and ppc64 supports 4K and 16M. A TLB is a cache of virtual-to-physical translations. Typically this is a very scarce resource on processor. Operating systems try to make best use of limited number of TLB resources. This optimization is more critical now as bigger and bigger physical memories (several GBs) are more readily available. https://www.kernel.org/doc/> Documentation/vm/hugetlbpage.txt

How to configure huge pages​

clyso@compute-21:~$ grep Hugepagesize /proc/meminfo
Hugepagesize: 2048 kB
clyso@compute-21:~$

echo 1024 > /proc/sys/vm/nr_hugepages

echo "vm.nr_hugepages=1024" > /etc/sysctl.d/hugepages.conf

total huge pages

clyso@compute-21:/etc/sysctl.d# grep HugePages_Total /proc/meminfo
HugePages_Total: 1024
clyso@compute-21:/etc/sysctl.d#

free hugepages

clyso@compute-21:/etc/sysctl.d# grep HugePages_Free /proc/meminfo
HugePages_Free: 1024
clyso@compute-21:/etc/sysctl.d#

free memory

clyso@compute-21:/etc/sysctl.d# grep MemFree /proc/meminfo
MemFree: 765177380 kB
clyso@compute-21:/etc/sysctl.d#

How to make huge pages available in kubernetes?​

restart kubernetes kublet on worker node​

sudo systemctl restart kubelet.service

verify in kubernetes​

Allocated resources​

clyso@compute-21:~$ kubectl describe node compute-21 | grep -A 8 "Allocated resources:"
Allocated resources:
  (Total limits may be over 100 percent, i.e., overcommitted.)
  Resource           Requests      Limits
  --------           --------      ------
  cpu                4950m (10%)   15550m (32%)
  memory             27986Mi (3%)  292670Mi (37%)
  ephemeral-storage  0 (0%)        0 (0%)
  hugepages-1Gi      0 (0%)        0 (0%)
  hugepages-2Mi      400Mi (19%)   400Mi (19%)
clyso@compute-21:~$

Capacity​

clyso@compute-21:~$ kubectl describe node compute-21 | grep -A 13 "Capacity:"
Capacity:
  cpu:                48
  ephemeral-storage:  1536640244Ki
  hugepages-1Gi:      0
  hugepages-2Mi:      2Gi
  memory:             792289900Ki
  pods:               110
Allocatable:
  cpu:                48
  ephemeral-storage:  1416167646526
  hugepages-1Gi:      0
  hugepages-2Mi:      2Gi
  memory:             790090348Ki
  pods:               110
clyso@compute-21:~$

Sources:

Manage HugePages
Brief summary of hugetlbpage support in the Linux kernel
Configuring Huge Pages in Red Hat Enterprise Linux 4 or 5

the mysql database was running out of space and we have to increase the pvc size for it.

nothing simple than that. just verified if we have enough free space left.

edited the pvc defintion of the mysql statefulset and set the spec storage size to 20 Gi.

a few seconds later the mysql database space was doubled.

**ceph version 15.2.8

**ceph-csi version: 3.2.1

kubectl edit pvc data-mysql-0 -n mysql

We had the problem of getting the correct authorizations for the Ceph CSI user on the pools.

We then found the following bug for the version prior to 14.2.12.

https://github.com/ceph/ceph/pull/36413/files#diff-1ad4853f970880c78ea0e52c81e621b4

Was then solved with version 14.2.12.

https://tracker.ceph.com/issues/46321