[Gdal-dev] Failed to acquire IOMutex in WarpRegion()

[Matt Wilkie]

> When you use -multi, the warper attempts to keep two threads working, one
> doing read/write, and one doing resampling in memory.  For synchronization
> purposes they use mutexes, and with a default timeout of 600 seconds.  If
> one thread waits more than 600s you will get the above error, and a warp
> failure.

Thank you Frank. It feels so much better when one knows what's
happening, even if it doesn't actually change the circumstances. :)

> Perhaps I ought to increase the timeouts quite a lot.  If you file a ticket,
> I'll look into it.  But from your side you can reduce the chunk size so it
> won't take so long to process.  Use -wm 100 instead of -wm 512 for instance.

I'm trying without -wm at all at the moment for testing, but I think
there must be something wrong with the computer. No matter what I do I
can't get the cpu usage to go to more than 10 or 15%, even photoshop
opening a 1gb image doesn't use more than 2%, ditto for gdalwarp
processing two 1gb images (without resampling mind you). I think in
this instance the timeout is helping diagnose a problem outside of
gdal so if no one else is reporting mutex errors I would leave it at
600 seconds for now.

-- 
-matt