[PATCH] Retry if NFS gives NOENT or EIO

From: Dan Christian <dchristian_at_google.com>
Date: Wed, 13 Feb 2008 16:15:39 -0800

In theory, this patch isn't needed. In practice it may be a good idea.

Some NFS implementations (e.g. linux-2.6 client with netapp server)
can return NOENT or EIO on files that have been renamed (e.g.
db/current). This is a NFS bug and only happens under load, but it
still might be wise to apply the work around.

[[[
* subversion/libsvn_fs_fs/fs_fs.c
Work around buggy NFS implementations.
In addition to retrying on ESTALE, also retry on NOENT and EIO
]]]

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe_at_subversion.tigris.org
For additional commands, e-mail: dev-help_at_subversion.tigris.org

text/plain attachment: noent2-patch6.txt

Received on 2008-02-14 01:15:55 CET

This message: [ Message body ]
Next message: Bert Huijben \(TCG\): "Messages and protocol crash on svn log with multiple targets"
Previous message: Eric Gillespie: "Re: svn commit: r29313 - trunk/subversion/bindings/swig/ruby/test"
Next in thread: Eric Gillespie: "Re: [PATCH] Retry if NFS gives NOENT or EIO"
Reply: Eric Gillespie: "Re: [PATCH] Retry if NFS gives NOENT or EIO"
Reply: Dan Christian: "Re: [PATCH] Retry if NFS gives NOENT or EIO"

Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]