On 11/9/21 10:49, Marc wrote:
I did an upgrade from 14.2.23 to 16.2.6 not knowing that the current
minor version had this nasty bug! [1] [2]
I'm sorry you hit this bug. We tried to warn users through
documentation. Apparently this is not enough and other ways of informing
operators about such (rare) incidents might be worthwile. I have put
this topic on the agenda for the upcoming user + dev monthly meeting
[1].
This is to be expected not? If you aim to get a broader audience to use ceph and start making tools like ceph-adm for just clicking next-next-next-done. You cannot expect then also that this audience is reading anything. I think this is mutually exclusive behaviour. So you are left only with not allowing the update to be done via such tools.
I do not know if Ceph has any formal documentation on how you should
upgrade your cluster. Most of us know (potentially learned the hard way)
it can pay off to read release notes and mailing lists. Regardless of
how you upgrade your cluster, i.e. by means of cephadm, ceph-ansible, by
hand or any other method it might be useful to get information in a
place where you, as an operator, are most likely to see it. And act
accordingly. I agree that tools like cephadm take the responsibility of
upgrading your cluster in a safe manner. And therefore it might make
sense to have it check if it is safe to do so. Or inform a user it
should upgrade ASAP because the current running release is at risk that
can be mitigated by upgrading. Ideally a general solution can be devised
that can be used by tools and humans alike.
Gr. Stefan
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx
