Some experience in using Zotero after Mendeley.
Have used Mendeley for several years, but recently had to migrate to a Linux workstation and Mendeley turned out to be a bit problematic in running on Linux and unstable in integrating with LibreOffice, especially when other people in the lab use Windows and other reference managers. For now I don't want to set up a Windows virtual machine only for running Mendeley with Microsoft Word.
Besides, I very much doubt that there is bright future ahead for Mendeley with Elsevier.
So I started looking for alternatives. OS is Linux Mint MATE and LibreOffice Write is my main paperwork program.
I used Zotero Standalone with Zotfile and Better Bibtext addons. Imported to Zotero with Menotexport.
Experience so far:
0. Not all of it is intrinsically tied to Zotero itself, but mostly to Menotexport, so don't take too much offence. Anyway, it should definitely be useful for other migrants.
Imported only one Mendeley folder (with a couple of dozens subfolders nested in two levels) out of my collection, since both the export and import were taking forever. Rearranging all of the subfolders after import into their proper structure was a bit tedious, Zotero was also nagging a lot about duplicates, since a lot of entries in Mendeley were sorted in several subfolders at the same time and import-export process turned them into full copies. So I had to manually remove the duplicates and then sort them back in multiple relevant subfolders. It also messed up entries that had multiple files tied to them, like the paper itself and supplementaries. So had to manually rebuild them too. Entries with missing dates got dated 0000.
1. Citations finally work. Even after resaving the file in MS Word. That's definitely a big plus over Mendeley.
It also doesn't beg for server connection.
2. Interface.
Left column: In Mendeley it highlights all the folders where currently selected entry is listed. It's convenient. It doesn't do this in Zotero.
Central: It remembers preferred sorting order! And you can even set secondary one! Mendeley has only one and it resets every launch to randomness for some reason. Being able to select which data columns to display is also nice, though it's not useful for me (name, author name, year, journal and add date covers all the needs).
Before working with Zotero, I never realized how elegant are two lines for displaying entries in Mendeley. It allows to fit far more information about the entry than one-line listing in Zotero.
Right column: It's bare and cluttered at the same time. There are too much empty, never to be used metadata entries, yet there is no place for PMID. The abstract is either fully closed and unreadable or fully open and takes up too much space, no middle ground.
Also in Mendeley it's very convenient when you can autofill the rest of metadata in one click on PMID, DOI or ArxiV ID. Inability to do this here is especially frustrating when apparently Zotero can add a new entry by PMID or DOI, but not complete an existing one.
Tags appear to be more useful than in Mendeley, where it's necessary to almost type them in manually for every document and thus they are pointless there. Unfortunately, here, in Zotero, they are made useless here by presence of countless autogenerated ones, existing even when the relevant option to create them is unflagged. And it's impossible to remove them (unless you're a masochist and willing to do this one by one).
In general the interface generally feels noticeably more sluggish than in Mendeley.
3. File operations and syncing.
I used to add new entries by making the program fetch the PDFs from a special download directory, and as I understand it, it's a function of Zotfile to perform such operation. Unfortunately, I was unable to make it work consistently. Sometimes a popup window appeared asking if I want to add the file (is there a way to make it stop asking the conformation; I already put the file in the special folder, of course I do want to import it!). Sometimes it didn't. Sometimes it combined several different PDFs into one entry.
Sometimes, especially with the an file, the metadata is not extracted correctly, and it's a bit frustrating that I had to perform extraneous actions with creating a new entry by DOI and then merging it with file-containing one, instead of updating original one in one click.
I use WebDAV-supporting site to sync the files and there are no problems with setting it up and uploading/downloading them. Unfortunately it's incompatible with storing all the PDFs in a clearly-structured folder by Zotfile.
4. One bug I noticed so far is that Zotero say that update is available, but user doesn't have the permission to write to it's dir. While Zotero was placed. at ~/zotero/, so there shouldn't be any problems.

So this is my experiences and thoughts so far. Maybe other users or even developers might find something useful here. Maybe someone could even offer some advice to mitigate the problems.