scaleio version help

Hi,

From what I see in the repository, there is a SDC for this kernel version under 2.0.6035.0 - please check out the following link:

ftp://ftp.emc.com/Ubuntu/2.0.6035.0/

Thank you,

Pawel

Hi Pawel,

the ftp requires authentication, I think that community users cannot access to these resources.

@Himanshu: yesterday I replied to your question giving you instructions to downgrade to kernel version to 4.4.0-15. I'm working that way to test ScaleIO v.2.0.0.1 (the version available for community users). It works great but I'd like, as you, to get an upgraded version of ScaleIO compatible with kernel version 4.4.0-34.

Regards,

Davide

Hi Davide,

Full FTP URL - with the authentication details - should be in ScaleIO Deployment Guide, can you please check if  you can access it?

Thanks,

Pawel

Hi Pawel, Davide,

Thanks for your responses. I am able to get service started by getting the correct driver module for my kernel and SDC version, from the EMC repository. The path and credentials are given in the deployment guide.

I am now stuck in another issue with scaleio, so overall it is not working. But my current issue is solved.

regards

Himanshu

Hi Pawel,

I didn't find the authentication details in ScaleIO Deployment Guide but thanks to your informations I realized that some way the ftp is publicly accessible so I made some google search and I found the correct information to access ftp.emc.com in puppet module guide here:

https://github.com/emccode/puppet-scaleio/blob/master/README.md

So if someone needs to access ftp in order to get scini and xcache updated modules for Ubuntu or CoreOS that is the complete FTP link:

ftp://QNzgdxXix:Aw3wFAwAq3@ftp.emc.com/

@Pawel: Thanks a lot for the informations, now I can keep my Ubuntu kernel updated in my ScaleIO test lab.

Davide

No problem Davide - the FTP URL you pasted is exactly the one from the Deployment Guide and as you can see it contains username and password in the URL string :-)

Pawel

Hi Pawel,

I noticed that the URL I posted contains credentials. I finally found the correct Deployment Guide you were referring to that contains detailed informations for Ubuntu deployment.

I was downloading the "Deployment Guide" on the community site, clicking on the second link on the section screenshotted here:

The name of the file downloaded from that link is: h15148-emc-scaleio-deployment-guide.pdf

But you were referring to the Deployment Guide included in the ScaleIO downloadable package, the file name of this document is: ScaleIO_v2_Deployment_Guide_Rev_02.pdf

Both of them are referring to ScaleIO 2.0 but they are completely different documents. That's why I didn't find the informations you were referring to.

Now Ubuntu kernel on all my nodes are updated and I have the matching xcache and sdc modules

Thanks again,

Davide

Hi Davide,

are you using xcache on Ubuntu nodes? I cannot seem to find xcache on my scaleio servers, although i am sure i used the package during installation.

what does xcache do?

regards

Himanshu

Hi Himanshu,

ScaleIO architecturally provides two different level for write cache. Every cache level can be enabled or disabled from the administrator as preferred. L1 cache is RAM cache, L2 cache is Flash Storage Cache. As you can see you can enable both of them for every SDS. L1 and L2 cache are useful to speedup read operations because if the cache is hot, that means that the data you are accessing to is already present in the cache it is served directly from L1 cache (if present here) or from L2 cache (if present here). Obviosly if is not present in cache it is read directly from the storage array. This technologies are useful to speed up read operations and also to unload the array of physical disks. For example if you have a cache read hit of 60% the physical disk array serves directly only 40% so the physical disk performance are preserved for writes.

Obviusly if the physical disks are mechanical the results are impressive! You can boost read performance at SSD level or to RAM level for cached data.

xcache (xtremcache) is the module that gives ScaleIO the ability to provide L2 cache feature (caching for read workload on Flash Devices).

I was in your situation, because I was sure to have it installed but wasn't. I found the way to add it to an existing installation without reboot. I added in my topology file (CSV) the column RFCache and I set the value "Yes" for every SDS. In ScaleIO installer in the "Install" tab I passed my new topology file and selected from drop down menu: "Add to existing sys" then I clicked "Upload Installation CSV". Then I went on with installation: since the topology file is very similar to the one used for installation (I added only RFCache) only the xcache module is added. After installation you have the same problem you had with scini module (the version of module mismatch with the kernel version).

At this point you have to do exacly the same operations you did to update scini module (the directory that contains the files to make the update of the module on Ubuntu is /bin/emc/scaleio/xcache_sync).

As soon you have updated the module you can launch xcache daemon running the command: "/etc/init.d/xcache start"

If the module works in your processes you could see some processes called: [sfc_thread] (as you can see in the screenshot attached:

If you didn't use a topology file during installation of ScaleIO you can extract it from the "Maintain" tab of ScaleIO installer. After extraction from the "Maintain" section you have to customize the CSV adding the RFCache column. Furthermore if you have to extract the CSV using this procedure make sure to update the password in the resulting file (the password is not exported) and the value about OS of every node is lost (so you have if the OS is linux or windows for every node).

If you have some problem during procedure write me here, I'm happy to help you.

Davide

Hi Davide

thanks for lot for the detailed explanation. I appreciate it.

I was wondering if missing xcache is part of my problem. Looks like it is not.

I am having trouble attaching scaleio volumes to Openstack VMs. Something prevents scaleio to send a successful response towards nova-compute. I see another thread with someone with the exact same issue as mine, so I will follow up there.

Best of luck with your testing.

regards

Himanshu