Re: leaking lots of unreferenced inodes (pg_xlog files?), maybe after moving tables and indexes to tablespace on different volume

<p dir="ltr"><br /> On Mar 13, 2013 3:04 AM, "Tom Lane" <<a
href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>>wrote:<br /> ><br /> > Palle Girgensohn
<girgen@FreeBSD.org>writes:<br /> > > ... I got lots of space freed<br /> > > up, but it seems that
afterthat the disk usage grows linearly (it seems<br /> > > to leave many inodes unreferenced).<br /> ><br />
>Hm.  We've seen issues in the past with PG processes failing to close<br /> > no-longer-useful files promptly,
but...<br /> ><br /> > > Strange thing is I cannot find any open files.<br /> ><br /> > ... that
suggeststhere's something else going on.<br /> ><br /> > > The unreferenced inodes are almost exclusively
around16 MB in size, so<br /> > > i.e. they would most probably all be pg_xlog files.<br /> ><br /> > Have
yougot any sort of WAL archiving active, and if so maybe that's<br /> > holding onto WAL files?  Not that it's clear
howcome lsof wouldn't<br /> > tattle on an archiving process either.<br /> ><br /> > > Stopping postgresql
brieflydid not help, I tried that.<br /> ><br /> > That seems to point the finger at some non-postgres cause.  I
confess<br/> > I can't guess what.<br /> ><p dir="ltr">Yeah, unreferenced inodes with no open files, and only
discoverablewith fsck sounds like a filsystem bug to me. Particularly since it showed up just after a operating system
upgrade,and doesn't go away with a postgres restart... <p dir="ltr">/Magnus