Colleague presented poster of our shared work at conferences without my knowledge or express permission

Ultimately there isn't a single valid approach here. The appropriate level of response depends on several factors, including the field in which the work is being performed, the nature of the conference, the status of the work relative to publishing practices in the field, and so on.

For instance, if you're in a field where posters are peer-reviewed before a conference, then it's a much more significant issue than presenting a poster at, for instance, a Gordon Research Conference where posters are explicitly considered to be "for discussion only" and are not intended to be used for publication. As for the author order, some societies (for instance, the APS) expect that the poster (or talk) be given by the first author as listed at submission—whether or not this is fundamentally the way it would be listed in a paper.

Looking at the ethical issue—has the work been published already somewhere or otherwise public? If so, your co-author may have thought it was OK to present a poster on the material; after all, you've already consented to publishing it elsewhere. On the other hand, if this is the first time the work has been presented, then it absolutely is ethically wrong to present the work without your consent.

The level of your response, though, should be scaled according to both the level of the offense and your desire to continue the collaboration. If you're angry that it was done but still want to work together, a somewhat measured response (the "sternly worded letter" approach) is probably best. If this has poisoned your relationship to the point where you can't see continuing to work together—and you want to have your colleague punished for the transgression—then you should make that clear. But I would hope the opening salvo here should be open and forthright dialogue with your colleague rather than a report to an ethics committee.

(Also, this advice can't really be generalized to large-scale collaborations where it simply is not possible to obtain informed consent from every participant on every paper produced as part of the collaboration.)