There is a bug 6540349 in the Solaris pcfs module concerning long
filenames: "PCFS endlessly loops on reading dirs if encountering
LFN entry with LDIR_Ord == {0xc0|0x80|0x40}". Apparently a patch
for the bug is expected to be available in June and an IDR could
be made available before that if necessary.

The bug is causing problems with my sunray servers. When a
corrupted usb stick is inserted, disk-related processes (df,
nautilus, utdomount, etc) hang and, to recover, a server reboot
is required. This is very inconvenient.

I see that third party programs are available which I believe
could detect such corrupt filesystems and fix them.

For example, debian has dosfstools containing dosfsck:

|> dosfsck verifies the consistency of MS-DOS file systems and  optionally
|> tries  to  repair  them. The following file system problems can be cor-
|> rected (in this order):
|> [...]
|> -      Directories with a large number of bad  entries  (probably  cor-
|>        rupt). The directory can be deleted.
|> [...]
|> -      Bad file names. They can be renamed.

So an obvious interim workaround to the problem (may even be a
good thing to do in the long-term) is to run such a program when
a USB stick is inserted, after the format has been detected but
before the mount is performed.

I note that /etc/rmmount.conf exists and allows for a call to
an fsck program (although it is not clear if a custom program
can be easily invoked without the effort of making a .so wrapper).
However, I do not know if the sunray usb mounting procedure even
makes use of rmmount. If not, is there a suitable hook where an
fsck call could be made?


thanks,

-jonathan

-- 
 Jonathan H N Chin   | deputy computer | Newton Institute, Cambridge, UK
<[email protected]> | systems manager | tel/fax: +44 1223 767091/330508
_______________________________________________
SunRay-Users mailing list
[email protected]
http://www.filibeto.org/mailman/listinfo/sunray-users

Reply via email to