Create Replicated Pool behavior

classic Classic list List threaded Threaded
7 messages Options
Reply | Threaded
Open this post in threaded view
|

Create Replicated Pool behavior

jcalcote
Hi Yaguang,

When I create a replicated pool in the Manage Pool screen, I see a spinner for a few seconds, and then I'm returned to the Storage Pools table - my new pool is not there. About 3-5 minutes later, my new replicated pool shows up.

Is this expected behavior?

Thanks,
John
Reply | Threaded
Open this post in threaded view
|

Re: Create Replicated Pool behavior

lihaijing
jcalcote wrote
Hi Yaguang,

When I create a replicated pool in the Manage Pool screen, I see a spinner for a few seconds, and then I'm returned to the Storage Pools table - my new pool is not there. About 3-5 minutes later, my new replicated pool shows up.

Is this expected behavior?

Thanks,
John
I tested.When I create a replicated pool,I can see it immediately.But strange, after a few seconds, PG conut changed from 64 to 100(I have 3 osds).

Reply | Threaded
Open this post in threaded view
|

Re: Create Replicated Pool behavior

ywang19
Administrator
In reply to this post by jcalcote
the expected behavior should be when the spinner is disappeared, the new created pool should show up. it should be a bug.
Reply | Threaded
Open this post in threaded view
|

Re: Create Replicated Pool behavior

ywang19
Administrator
In reply to this post by lihaijing
if the PG number per OSD is too low, vsm will automatically tune it. recently, we are developing a feature which allows user to assign pg count and stop auto-tune.
Reply | Threaded
Open this post in threaded view
|

Re: Create Replicated Pool behavior

jcalcote
Yaguang,

Would you mind providing a set of options for the "create replicated pool" page that work for you? I'm not really a ceph expert, so I'm not entirely sure what values I should be using. I know these options are probably deployment dependent, but if you could just tell me what your test deployment looks like and what options you use, it would be helpful. Then I could try variations on your options that are appropriate for my deployment.

The problem is this may be the result of using numbers that are way off of normal, but I'm not sure what "normal" is for this situation.

Thanks,
John
Reply | Threaded
Open this post in threaded view
|

RE: Create Replicated Pool behavior

ywang19
Administrator

Hi john,

 

Here is a screen shot I used to create replicated pool, I can create it successfully.

 

 

 

-yaguang

 

From: jcalcote [via vsm-discuss] [mailto:ml-node+[hidden email]]
Sent: Tuesday, January 12, 2016 5:38 AM
To: Wang, Yaguang
Subject: Re: Create Replicated Pool behavior

 

Yaguang,

Would you mind providing a set of options for the "create replicated pool" page that work for you? I'm not really a ceph expert, so I'm not entirely sure what values I should be using. I know these options are probably deployment dependent, but if you could just tell me what your test deployment looks like and what options you use, it would be helpful. Then I could try variations on your options that are appropriate for my deployment.

The problem is this may be the result of using numbers that are way off of normal, but I'm not sure what "normal" is for this situation.

Thanks,
John


If you reply to this email, your message will be added to the discussion below:

http://vsm-discuss.33411.n7.nabble.com/Create-Replicated-Pool-behavior-tp356p364.html

To start a new topic under vsm-discuss, email [hidden email]
To unsubscribe from vsm-discuss, click here.
NAML

Reply | Threaded
Open this post in threaded view
|

RE: Create Replicated Pool behavior

jcalcote
Hi Yaguang,

Problem solved. I noticed an error in the controller vsm-api.log during attempt to execute the pool create command. The problem was that we're importing a crushmap that defines a ruleset called replicated_ruleset, but when we define our cluster.manifest, we were defining the storage group with a different name (high_performance). Thus, VSM was seeing two storage groups - one called high_performance, and the other called replicated_ruleset.

I fixed this by defining our cluster.manifest with storage groups matching the imported crush rulesets.

Thanks,
John