[ClusterLabs] attrd/attrd_updater asynchronous behavior

Jehan-Guillaume de Rorthais jgdr at dalibo.com
Mon Apr 16 22:12:05 UTC 2018


I got an answer on IRC from Ken Gaillot. Bellow his answer for tracking purpose.

On Mon, 16 Apr 2018 23:28:39 +0200
Jehan-Guillaume de Rorthais <jgdr at dalibo.com> wrote:
[...]
> * does looping until the value becomes available is enough to conclude all
>   other node have the same value? Or is available only locally on the action's
>   node and not yet "replicated" to other nodes? 

kgaillot: « 
  That issue has come up recently in a different context. you are
  correct, currently there is no guarantee that the value has been set
  anywhere, and looping until the query comes back ensures that the new value
  is in the local attrd only.

  The solution will probably be to offer something like a --wait option that
  doesn't return until the value is available (maybe locally, maybe everywhere,
  or maybe that's part of the option).
»

I'll fill a bz for tracking purpose of such feature as discussed on IRC.

> * any other suggestions about how we could share values synchronously with all
>   other nodes?

Any suggestion is very welcome...


More information about the Users mailing list