[
https://issues.apache.org/jira/browse/TINKERPOP-1337?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16366959#comment-16366959
]
stephen mallette commented on TINKERPOP-1337:
---------------------------------------------
I don't know if Gremlin Server should get this advanced. We're seeing more and
more implementations that are "server graphs" - DSE Graph, Neptune, CosmosDB,
etc. that only rely on the Gremlin Server protocol and not necessarily the
server itself. I think we should leave it to those graph providers to handle
this kind of stuff (if they feel that they need it) and Gremlin Server stay
focused on hosting embedded graphs. I will leave this open for a bit in case
others have comments, but I'm inclined to close it.
> Provide an "add jar" endpoint
> -----------------------------
>
> Key: TINKERPOP-1337
> URL: https://issues.apache.org/jira/browse/TINKERPOP-1337
> Project: TinkerPop
> Issue Type: Improvement
> Components: server
> Affects Versions: 3.2.0-incubating
> Reporter: Daniel Kuppitz
> Priority: Major
>
> Gremlin Server should provide something (an endpoint?) that allows the user
> to add new jar files, without the need to restart the server.
> We've talked about it before, but I thought it might be a good idea to have a
> ticket where we collect some thoughts.
> One particular problem we've talked about is this: What if someone wants to
> update a previously loaded jar? The initial loading of a new jar file seems
> to be a smaller problem; unloading a jar file to update it with a newer
> version seems to be a real problem. I would say we simply shouldn't support
> that. I've looked into other projects (e.g. Hive) and there're ways to load
> new jars, but not to unload them later. If you really need to get rid of a
> previously loaded jar, then you'll have to restart the server / JVM.
> Another problem I see are distributed environments, where you have multiple
> Gremlin Servers running (none knows about the existence of the others) that
> are requested in a round-robin fashion. I don't have a good idea on how to
> handle this problem, but a first step in the right direction may be to allow
> uploads of jar files to distributed file systems. Perhaps Gremlin Server
> instances could then monitor the contents of a predefined directory within
> the DFS.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
