In a "mdworker quit unexpectedly" loop after turning on Smart Sync

Kidhack

Ever since turning on Dropbox Smart Sync, Spotlight has been going nuts and using tons of CPU. The helper app "mdworker" is constantly crashing. Will report back after Dropbox finishes indexing all the Smart Sync files. 

PS. Funny you can't link Dropbox photos to the image uploader in this forum.

 

Ross_S

Hi everyone,

 

 I wanted to give you an update on the issue of mdworker crashing when reading a mismatch in the expected logical vs physical size of PDFs on Mac.

 

Firstly we've not forgotten about this, a lot of work has been going on in the background to resolve the issue. It has required extensive code re-writing. For those not familiar with the issue, users may see within the logs mdworker crashing. This crash is fairly silent, though will appear in logs. There's no risk to data, but for those who like to inspect clean logs this can cause a lot of noise and may be viewed as sub-optimal. 

 

Based on the feedback raised by yourselves on this forum we've spent considerable time working toward a fix. We've changed the way we write Smart Sync placeholders (ie the files that look like files, but take up 0kb until recalled) for those on 10.13.5 and later. The new sparse format will not trigger an mdworker crash. 

 

We've tested within the beta to validate the code and at this point have now rolled the fix / changed behavior out to 100%. 

 

This is a forward fix, so from now on all and any placeholders created will not suffer this issue. For those who have the placeholders (the empty files with the grey cloud symbol on them) already on their machines there are a few options:

 

1. Unlink and relink your Dropbox. Note: this will trigger a reindex of content that may take minutes to a few hours depending on your filecount. 

1. Remove the content, via Selective Sync, then return it. Similarly, but perhaps preferably, this will remove the placeholders then re-add them quickly and non-destructively.

There are a few other options, but those keen to avoid this error being highlighted in logs can take these steps.

 

Many thanks for your feedback on this issue here, it's driven a clear change in product. We pay close attention to the feedback surfaced via this channel so if you have other wishes please either add to existing threads, up-vote ideas, or create your own thread to spark discussion or development.