Double blinded review but the appendix contains author's name
When in doubt: yes. Let them know when and how you found this potentially compromising information. Especially if you think this may impact your objectivity (e.g. "I don't recognize this name, so I'm not going to trust him as much and may/will be more critical than I otherwise would"). But even if you think your evaluation will be unaltered, you should let the editor know. They need to be the one to decide if the review process can still meet the necessary criteria, if the authors should be notified and asked to alter the file to eliminate the metadata, etc.
In your case it sounds like you did the entire review under appropriate double blinding, fully unaware of the authors while making the evaluation, and then at the very end of the process you gleaned a potential clue about the author name. If so, then the editor may accept that no actual problem has arisen, and may simply try to take steps to ensure such metadata is no longer visible to reviewers.
Things get more complicated if there are multiple reviewers for this paper: it now becomes possible that one or more of them saw this data early in the process, and it potentially compromised their review, but yet this went unacknowledged. This makes it all the more important to inform the editor.
I have always informed the editor and in every case, he hasn't cared. I seem to have a super-power in that I can recognize people by their writing style. So I have always known who the author of the paper was, without any tell-tale flaws in data scrubbing. There's nothing the editor can do to thwart my super-power. (Well, perhaps he could have someone translate it into Chinese and then someone else translate it back to English, but it'd be better to just get a different referee.) I just put in my comments to the editor that, "I can tell this is Bobby Boogy's work." So I go for full disclosure and let the editor handle it as he will.
Unless you think that this might have impacted your objectivity: No, there is no need to actively involve the editor. As you revealed the identity only after writing the review ("before submitting") objectivity is probably not an issue.
It is important to understand that double-blind is far from being perfect: In many cases the "secret of authorship" just does not hold for long. You know your colleagues, what they are working on, which tools they typically employ, their style of writing and so on.
Hence, so even if there are some common rules, double-blind should not be understood in a too formal manner. It basically is a code of conduct. (1) The authors agree to try their best to hide their identity. However, there is also the other side: (2) The reviewers agree to not actively try to reveal the author's identity.
In this realm you as the reviewer are supposed to not look into the PDF metadata. If you stumble across a single, apparently accidentally left-over name, try to ignore it. Yes, all this should not happen if the authors had done their job. However, we are all humans and as long as neither your objectivity is impaired nor you have the impression that the authors tried to "drop" their identity, there is no need to make too much of a fuzz about it.