You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I would like to be able to specify an ID of the executor as a variable in the Data directory string. To provide the best use case example, it would be the kafka-mesos framework. There, I say I want 4 brokers, each is assigned an ID with stickyness. so broker1, broker2, broker3, broker4. I can update settings for that broker, I can also specify different data directories for each broker. This allows me to have a unified name space. (Say if I have a high performance NFS share) where I can have each elastic search executor have it's own directory, and no matter what mesos agent it starts on it always has access to ITS data directory. Now, a variable would work, or specify a data-dir for each executor. the kafka-mesos framework sees the scheduler as managing all the various brokers individually, where this framework seems to treat them all as equals...
Regardless, having the ability to have each executor have it's own data directory that is available to it at the same mount point would be handy (or maybe I can already do this?)
The text was updated successfully, but these errors were encountered:
I would like to be able to specify an ID of the executor as a variable in the Data directory string. To provide the best use case example, it would be the kafka-mesos framework. There, I say I want 4 brokers, each is assigned an ID with stickyness. so broker1, broker2, broker3, broker4. I can update settings for that broker, I can also specify different data directories for each broker. This allows me to have a unified name space. (Say if I have a high performance NFS share) where I can have each elastic search executor have it's own directory, and no matter what mesos agent it starts on it always has access to ITS data directory. Now, a variable would work, or specify a data-dir for each executor. the kafka-mesos framework sees the scheduler as managing all the various brokers individually, where this framework seems to treat them all as equals...
Regardless, having the ability to have each executor have it's own data directory that is available to it at the same mount point would be handy (or maybe I can already do this?)
The text was updated successfully, but these errors were encountered: