Re: rmtree() failure on Windows - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: rmtree() failure on Windows
Date
Msg-id 200410271516.i9RFGrc14743@candle.pha.pa.us
Whole thread Raw
In response to Re: rmtree() failure on Windows  (Andrew Dunstan <andrew@dunslane.net>)
List pgsql-hackers
Andrew Dunstan wrote:
> 
> 
> Reini Urban wrote:
> 
> >>>
> >>> 300 secs (!) fs timeout is really broken.
> >>> Looks more like a locking or network timeout issue.
> >>> What error codes does unlink(3) return?
> >>
> >
> >> success.
> >
> >
> > Oops! 5min timeout for success is certainly problematic.
> >
> >
> 
> You misunderstood. The 300 secs is not in waiting for unlink() to 
> return, it is in waiting for its effects to be seen by rmdir() (i.e. for 
> the entry to actually be cleared from the directory). unlink() is 
> returning very quickly. If the bgwriter is disabled then the entries are 
> cleared very quickly too (i.e. before we even get to rmdir())

What I am thinking is that the way we open these files allows us to
delete them while they are still open, but perhaps it doesn't allow us
to actually delete the directory that contains the files.

Also, it is possible bgwriter is keeping some files open in that
directory and that is causing the long delays.

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
359-1001+  If your life is a hard drive,     |  13 Roberts Road +  Christ can be your backup.        |  Newtown Square,
Pennsylvania19073
 


pgsql-hackers by date:

Previous
From: Oliver Elphick
Date:
Subject: Re: 8.0b4: COMMIT outside of a transaction echoes
Next
From: "Zeugswetter Andreas DAZ SD"
Date:
Subject: Re: rmtree() failure on Windows