Since it's possible to close a question as a duplicate of multiple other questions and thus provide a solution to each one of the multiple problems apparent in it*, closing a question about a complex package management issue as a duplicate sometimes requires a collaborative effort beyond clicking along with the previous reviewer.
Let's say someone with close-vote or flagging privilege perceives that multiple existing posts will solve all the problems in the question. Unless that person happens to be a mod or holds a gold badge in any of the post's existing tags (unfortunately you can't give yourself powerz by editing in the tag) they can only vote to close the post as a duplicate of one question, so they have to make a decision on which question to choose. If there is any difference in the level of severity in the problems the OP is having, I guess that first voter should generally try to choose the dupe target that solves the most critical problem. They can then add another comment or edit the automatic possible duplicate comment to add links to other dupe targets so that subsequent reviewers can add votes to close against those targets, so that the final banner solves all the problems, from the most severe to the least.
However, if the first voter has voted to close the question as a duplicate of one that resolves only a non-critical problem and not left any comments about additional possible duplicates, the subsequent reviewers are left with the task of finding the other necessary dupe targets. This is fine if the other required targets are well-known canonical posts like Permission denied, are you root? or Unable to lock the administration directory. Is another process using it? and no experienced reviewer will disagree that the question is a dupe of those targets. In such a case, the first voter probably did the right thing by choosing the most obscure of the required targets, leaving subsequent reviewers whose fingers know the canonical posts by muscle memory to effortlessly drop in the top search result.
However, if there are no duplicate targets to cover the critical parts of the question, then imho it's not a good idea to close the post as a duplicate of a post that only resolves the non-critical part. If someone has suggested such a duplicate, and you aren't aware of targets to cover the other issues in the question, then as a reviewer you might want to
- Ask the person who suggested the duplicate if they have other targets in mind to cover the more critical parts of the question (to engage in collaborative multi-target closing)
- Point out for the benefit of subsequent reviewers that this target only solves non-critical problems, leaving a more fatal problem unresolved, so the question either needs an answer (and so should not be closed, since closing prevents answering(!)), or more dupe targets, or if it's not answerable within scope should be closed for another reason.
- Do what you would do anyway, for example:
- If the question already has a good/upvoted/accepted answer and is appropriate for the site, vote to leave open.
- If the question is appropriate for the site, if it is new or the OP is active or the question is currently answerable, then vote to leave open and answer or seek an answer.
- If the question lacks sufficient information, vote to close as unclear, leaving a comment on what's needed if possible.
- If comments indicate the problem is not reproducible for the OP or they have apparently abandoned the question, and it's not a common problem that others will be able to reproduce either (if that were the case there would probably be a dupe target), vote to close as no-repro.
- etc etc for other close reasons and possible actions...
* multiple problems? Shouldn't the question be closed as too broad? Probably not, because a single cause (such as mixing repositories, or a botched effort to add repositories) can result in multiple different error messages, and because when people ask questions about their broken package management system, we ask them to give the full verbatim output of sudo apt update and/or sudo apt install -f and so on, to give a holistic picture of the trainwreck that has mysteriously been caused by the antics of OP's cat/3 year old/prank-loving colleague. Chopping that output up and boxing it into a series of questions wouldn't help in most cases. Multiple dupe targeting is really helpful in these situations when used correctly, since OP can be directed to comprehensive resources without someone duplicating large quantities of community effort in a single long answer that covers all the errors we have seen a hundred times.
That said, important to note here, I really don't want to discourage answering. Answers are what actually help people in the end. Dupe targets are only as useful as their answers. If a question seems not to be a duplicate after all or you think it would benefit more from a direct answer than from being closed against some similar thing(s), then please consider answering it if you can.
