From: Timothy Shimmin <tes@sgi.com>
To: Richard Hartmann <richih.mailinglist@gmail.com>
Cc: Nathan Scott <nscott@aconex.com>, Bill Kendall <wkendall@sgi.com>,
xfs@oss.sgi.com
Subject: Re: xfsrestore over ssh?
Date: Wed, 23 Jul 2008 10:19:24 +1000 [thread overview]
Message-ID: <4886790C.7040100@sgi.com> (raw)
In-Reply-To: <2d460de70807220351i7c468de5o33fa5aa5e9ddab99@mail.gmail.com>
Richard Hartmann wrote:
> On Tue, Jul 22, 2008 at 09:04, Timothy Shimmin <tes@sgi.com> wrote:
>> If you could strace it then we could see what call was failing> in path_to_fshandle().
> First with -ttt to get rid of a tty allocation error, then without:
> % cat root.xfs | ssh -ttt root@192.168.1.213 strace xfsrestore - /mnt/sda3
>From your output:
> open("/mnt/sda3", O_RDONLY|O_LARGEFILE) = 4
> ioctl(4, 0xc01c5868, 0xbfd4178c) = -1 EINVAL (Invalid argument)
> close(4)
(I probably should have just asked for:
# strace -o output -e trace=open,ioctl xfsrestore ...)
> On Tue, Jul 22, 2008 at 12:49, Richard Hartmann
> <richih.mailinglist@gmail.com> wrote:
>
>> > The Knoppix boot CD is am using to restore has 2.2.24-1, btw.
>
> Grml to the rescue.. I downloaded a grml iso, which has 2.2.45-1,
> and was able to restore the image.
>
> Extremely annoying, as I spent quite some time poking various
> options, thinking I had been doing something very basic very
> wrong. I assume the issue has been fixed between .24 and .45.
>
>
> Thanks again,
> Richard
Okay, my best guess is that you are trying to restore to a non-xfs
filesystem. This was fixed again in 2.2.26 so that it would work
on a non-xfs filesystem.
Otherwise the xfs ioctl will be being called and get back an EINVAL
as we saw in the strace.
Personally, I can't remember why we are bothering with handle code
in restore (in dump, yes, because of bulkstat but restore....
I vaguely remember a comment about optimitisation).
Looking at the code (annotated version) in restore:
wkendall |1.34 |master-melb:xfs-cmds:187436a| /* effectively initialize libhandle on this filesystem by
wkendall |1.34 |master-melb:xfs-cmds:187436a| * allocating a file system handle. this needs to be done
wkendall |1.34 |master-melb:xfs-cmds:187436a| * before any open_by_handle() calls (and possibly other
wkendall |1.34 |master-melb:xfs-cmds:187436a| * libhandle calls).
wkendall |1.34 |master-melb:xfs-cmds:187436a| */
wkendall |1.34 |master-melb:xfs-cmds:187436a| if ( persp->a.dstdirisxfspr ) {
wkendall |1.34 |master-melb:xfs-cmds:187436a| void *fshanp;
wkendall |1.34 |master-melb:xfs-cmds:187436a| size_t fshlen=0;
wkendall |1.34 |master-melb:xfs-cmds:187436a|
wkendall |1.34 |master-melb:xfs-cmds:187436a| if(path_to_fshandle(persp->a.dstdir, &fshanp, &fshlen)) {
wkendall |1.34 |master-melb:xfs-cmds:187436a| mlog( MLOG_NORMAL,
wkendall |1.34 |master-melb:xfs-cmds:187436a| _("unable to construct a file "
wkendall |1.34 |master-melb:xfs-cmds:187436a| "system handle for %s: %s\n"),
wkendall |1.34 |master-melb:xfs-cmds:187436a| persp->a.dstdir,
wkendall |1.34 |master-melb:xfs-cmds:187436a| strerror( errno ));
wkendall |1.34 |master-melb:xfs-cmds:187436a| return BOOL_FALSE;
wkendall |1.34 |master-melb:xfs-cmds:187436a| }
wkendall |1.34 |master-melb:xfs-cmds:187436a| /* libhandle has it cached, release this copy */
wkendall |1.34 |master-melb:xfs-cmds:187436a| free_handle(fshanp, fshlen);
wkendall |1.34 |master-melb:xfs-cmds:187436a| }
wkendall |1.34 |master-melb:xfs-cmds:187436a|
----------------------------
revision 1.34
date: 2005/02/10 02:14:17; author: wkendall.bonnie.engr.sgi.com; state: Exp; lines: +34 -26
modid: master-melb:xfs-cmds:187436a
Fix a couple of spots in xfsrestore where XFS-specific calls were
being made without first checking to see if the filesystem being
restored is XFS.
Merge of master:xfs-cmds:187436a by kenmcd.
----------------------------
xfsdump-2.2.26 (09 February 2005)
- Fix xfsrestore so that it can restore to non-XFS filesystems again.
--Tim
next prev parent reply other threads:[~2008-07-23 0:18 UTC|newest]
Thread overview: 16+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-07-21 20:17 xfsrestore over ssh? Richard Hartmann
2008-07-21 20:33 ` Bill Kendall
2008-07-22 3:02 ` Richard Hartmann
2008-07-22 3:59 ` Timothy Shimmin
2008-07-22 4:43 ` Richard Hartmann
2008-07-22 6:20 ` Donald Douwsma
2008-07-22 10:40 ` Richard Hartmann
2008-07-22 6:33 ` Nathan Scott
2008-07-22 7:04 ` Timothy Shimmin
2008-07-22 10:51 ` Richard Hartmann
2008-07-22 10:52 ` Richard Hartmann
2008-07-23 0:19 ` Timothy Shimmin [this message]
2008-07-23 11:39 ` Richard Hartmann
2008-07-22 10:45 ` Richard Hartmann
2008-07-22 10:49 ` Richard Hartmann
2008-07-22 12:00 ` Richard Hartmann
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4886790C.7040100@sgi.com \
--to=tes@sgi.com \
--cc=nscott@aconex.com \
--cc=richih.mailinglist@gmail.com \
--cc=wkendall@sgi.com \
--cc=xfs@oss.sgi.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox