At most, yes.
The properties which affect both would be the ones which start with
"general." or "instance.", with the latter being ones which must be the
same across the cluster in order for servers to participate in the same
cluster.
For all X, not in {general, instance}, properties starting with "X." should
only affect servers of type X. Otherwise, that's almost certainly a bug.
On Tue, Oct 4, 2016 at 9:21 PM Jeff Kubina <[email protected]> wrote:
> So just to clarify, changing a tserver.* option would at most only
> require a restart of all the tservers, not a restart of the master?
>
>
> On Tue, Oct 4, 2016 at 9:14 PM, Christopher <[email protected]> wrote:
>
> Right now, I think you'd probably have to track down where that particular
> property is used in the code to determine its lifecycle. I think it's going
> to take some work to wrangle these into discrete sets for documentation
> purposes, in the shell or otherwise. Some properties are only used during
> certain times early in the server's lifecycle. Other properties are used on
> demand. Some of those on demand properties are probably cached into
> internal state for indefinite periods of time. It's hard to say which are
> which without investigating each property individually (or through
> empirical testing).
>
> On Tue, Oct 4, 2016 at 9:04 PM Jeff Kubina <[email protected]> wrote:
>
> That would be very helpful, but a note in the documentation would be fine
> initially. Is there an easy way to determine this from the source code?
>
> --
> Jeff Kubina
> 410-988-4436 <(410)%20988-4436>
>
>
> On Tue, Oct 4, 2016 at 8:59 PM, Christopher <[email protected]> wrote:
>
> Some do, some don't. One thing we could add to the shell is a notification
> that a restart is necessary for a particular change. Possibly.
>
> On Tue, Oct 4, 2016, 20:25 Dave <[email protected]> wrote:
>
> I don't think so.
>
> On Oct 4, 2016 8:21 PM, Jeff Kubina <[email protected]> wrote:
>
> Does changing the values of tserver configs in the accumulo shell, like
> "config -s tserver.server.threads.minimum=256", require a restart of all
> the tservers to become effective?
>
>
>
>
>
