[RFE] configuration option to tune oplog size
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Triaged
|
Low
|
Unassigned |
Bug Description
Sometimes users don't enable-ha at the beginning of their journey with juju, when they decide to do it is possible that the database has gotten too big (logs, many models, images, etc) and mongodb could get stuck when replicating the data to the new node, other situation in which a small oplog size could affect a juju controller is when one of the replicas is powered off for maintenance.
Now that the leadership logic is being moved out of mongodb the writes are less, without the update-status changes in the database the writes are even less, but still a busy environment with DEBUG level set for the units and controller(s) may produce a considerable amount of writes.
References:
https:/
https:/
https:/
Changed in juju: | |
status: | New → Triaged |
importance: | Undecided → Medium |
We do already have agent.conf "MONGO_OPLOG_SIZE", which I thought we
exposed in controller config. IIRC, we don't let you change it after
bootstrap, as there are some sort of issue around it with mongo (I think
the issue is that you can't change it without restarting mongo, I don't
remember all the details).
But we do have a flag for agent.conf, I don't quite see a way that it is
exposed to the user.
On Mon, Sep 10, 2018 at 4:13 PM, Richard Harding <<email address hidden>
> wrote:
> ** Changed in: juju /bugs.launchpad .net/bugs/ 1790920 /bugs.launchpad .net/juju/ +bug/1790920/ +subscriptions
> Status: New => Triaged
>
> ** Changed in: juju
> Importance: Undecided => Medium
>
> --
> You received this bug notification because you are subscribed to juju.
> Matching subscriptions: juju bugs
> https:/
>
> Title:
> [RFE] configuration option to tune oplog size
>
> To manage notifications about this bug go to:
> https:/
>