Fault tolerance system

Rajesh Bhardwaj

unread,

Mar 5, 2016, 6:56:30 PM3/5/16

to Sparksee

Hello,

Will it be possible to create a high availability production ready system based on sparksee.

From what i understood (from documentation) certain key features like rollback is missing. Any updates with regards to that

What can be workaround for that (ours is both write and read intensive operation , we need active master slave configuration)

regards

c3po.ac

unread,

Mar 8, 2016, 6:49:26 AM3/8/16

to Sparksee

Hello,

Our HA system current version is thought mainly for read horizontal scaling, you must take into account that writes are always redirected to the master whileas the slaves can only perform reads and record the master changes. Future versions of Sparksee's HA must include the reassignment of master but unfortunately we still can't share a date for the release that includes this improvement.

We do offer Rollback on Sparksee's transactions, check here in the documentation to read more about it: http://sparsity-technologies.com/UserManual/API.html#transactions

Thanks,

El diumenge, 6 març de 2016 0:56:30 UTC+1, Rajesh Bhardwaj va escriure:

Rajesh Bhardwaj

unread,

Mar 9, 2016, 11:23:14 AM3/9/16

to Sparksee

Hello,

My requirement is a fault tolerant system, I am not looking for scaling. If the master goes down, will slave automatically click in ?

and then when master again comes back on, will it be able to sync with current master

regards

sparkseegdb

unread,

Mar 10, 2016, 5:45:48 AM3/10/16

to Sparksee

Hello Rajesh,

In the current version of Sparksee's HA functionality if the slave nodes fails, it is not a problem at all, you are going to be able to work and when it comes up it will sync with the master, but the same thing does not apply if the master fails. Since the master is the one in charge of the writes and sync if it fails it won't be re-assigned and thus the system can't synchronize.