xfs_repair said "cache_purge: shake on cache 0x80f5288 left 1 inodes!?"

xfs_repair said "cache_purge: shake on cache 0x80f5288 left 1 inodes!?"

[Håkan Lindqvist]

[-- Attachment #1: Type: text/plain, Size: 692 bytes --]

Hi!

When I ran xfs_repair (2.8.11) to fix a filesystem that had been bitten
by the Linux 2.6.17.<7 bug the following happened:

All seemed to go well, xfs_repair found the problematic directory inode
and rebuilded it.

At the end (right before printing "done"), it did however say
"cache_purge: shake on cache 0x80f5288 left 1 inodes!?" thrice.

The filesystem appears to be fine so far (and the directory that
previously caused the filesystem to shut down is ok again), but that
error/warning message sounded a little scary.

Is this problem harmless or should I be worried?


Regards,
Håkan Lindqvist

(Please CC me on any replies as I'm not subscribed to the list.)

[-- Attachment #2: smime.p7s --]
[-- Type: application/x-pkcs7-signature, Size: 3165 bytes --]

RE: xfs_repair said "cache_purge: shake on cache 0x80f5288 left 1inodes!?"

[Barry Naujok]

Hi, 

> -----Original Message-----
> From: xfs-bounce@oss.sgi.com [mailto:xfs-bounce@oss.sgi.com] 
> On Behalf Of Håkan Lindqvist
> Sent: Friday, 29 September 2006 1:49 AM
> To: xfs@oss.sgi.com
> Subject: xfs_repair said "cache_purge: shake on cache 
> 0x80f5288 left 1inodes!?"
> 
> Hi!
> 
> When I ran xfs_repair (2.8.11) to fix a filesystem that had 
> been bitten
> by the Linux 2.6.17.<7 bug the following happened:
> 
> All seemed to go well, xfs_repair found the problematic 
> directory inode
> and rebuilded it.
> 
> At the end (right before printing "done"), it did however say
> "cache_purge: shake on cache 0x80f5288 left 1 inodes!?" thrice.
> 
> The filesystem appears to be fine so far (and the directory that
> previously caused the filesystem to shut down is ok again), but that
> error/warning message sounded a little scary.
> 
> Is this problem harmless or should I be worried?

This cache_purge comment is harmless. It's informing us that some buffers
have reference counts remaining that weren't unreferenced when read. All
dirty buffers do get written.

Barry.