Issue 219 in ganeti: ganeti-rapi should be optional

[gan...@googlecode.com]

Status: New
Owner: ----

New issue 219 by iust...@gmail.com: ganeti-rapi should be optional
http://code.google.com/p/ganeti/issues/detail?id=219

Just thought about this: on a small cluster or a special deployment, we
might not want to control ganeti from outside. As such, it should be
possible to disable rapi at configure time *or* runtime, like we do for
confd.

Ideally this and confd should be changeable at runtime, not only at
configure time, as that prevents distributions from offering both.

[gan...@googlecode.com]

Updates:
Owner: ultrot...@google.com

Comment #2 on issue 219 by ultrot...@google.com: ganeti-rapi should be
optional
http://code.google.com/p/ganeti/issues/detail?id=219

This is already possible, I believe, by modifying the defaults file. I
think it just needs documentation. (no point in adding an extra parameter,
if we can do without, right?)

[gan...@googlecode.com]

Comment #3 on issue 219 by ius...@google.com: ganeti-rapi should be optional
http://code.google.com/p/ganeti/issues/detail?id=219

No, it's not just that. The code assumes that a healthy cluster means RAPI
is running on the master. Hence watcher will restart RAPI if dead, etc.

[gan...@googlecode.com]

Comment #4 on issue 219 by ultrot...@google.com: ganeti-rapi should be
optional
http://code.google.com/p/ganeti/issues/detail?id=219

Which is not a big deal, if it restarts /bin/true, right? :)
But of course, an extra way to make it not do it is nicer.

Thanks,

Guido


--
You received this message because this project is configured to send all
issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

[gan...@googlecode.com]

Updates:
Status: Accepted
Owner: ---
Labels: Milestone-Release2.9

Comment #5 on issue 219 by ultrot...@google.com: ganeti-rapi should be
optional
http://code.google.com/p/ganeti/issues/detail?id=219

(No comment was entered for this change.)