VirtualBox

Ticket #7791 (new defect)

Opened 3 years ago

Last modified 3 years ago

If disk images are on a shared storage(eg. samba), clients crashes if there is networking error.

Reported by: yiyihu Owned by:
Priority: major Component: virtual disk
Version: VirtualBox 3.2.12 Keywords:
Cc: Guest type: other
Host type: Windows

Description

I keep all disk images in a samba share, and sometimes, If there is network problem, The guests fail constantly, Even if the share which contains disk images offline and go back within in a second, The guest fails(both windows and linux) still. Asked in #vbox channel, And finally the conclusion might be:

00:13 < lacqui> But honestly, the issue is that VBox is attempting to access a file that it assumes is constantly available 00:13 < lacqui> When there are network issues, that may no longer be the case. 00:13 < lacqui> VBox has the file open constantly. 00:13 < lacqui> I beleive that file pointers may be lost (on the OS level); I don't know if there is an application-level method of bypassing that 00:14 < xinming> Since the samba share returns again really soon, So I believe there may be a work around to detect this, and re-open the file.

Possible to fix this? Or if this won't be fixed, Is there a way to prevent this situation please?

Thanks!

Attachments

cifs test-2010-12-06-22-33-01.log Download (38.6 KB) - added by yiyihu 3 years ago.
The log for running vbox, while restart the samba server.
cifs test-2010-12-06-23-10-41.log Download (170.3 KB) - added by yiyihu 3 years ago.
The log for running vbox, while restart the samba server.

Change History

Changed 3 years ago by yiyihu

The log for running vbox, while restart the samba server.

Changed 3 years ago by yiyihu

The log for running vbox, while restart the samba server.

comment:1 Changed 3 years ago by yiyihu

The "cifs test-2010-12-06-22-33-01.log" is the old one which is on samba server too, Where vbox failed to log when samba restarts.

The latter "cifs test-2010-12-06-23-10-41.log" is the one where I store the image on the samba server, and keep log on local machine. So that the whole right log is captured.

comment:2 Changed 3 years ago by klaus

Dealing with such frequent CIFS failures for disk images might be doable, in the same way as with iSCSI, essentially by pausing the VM, since no one knows when your flaky CIFS server comes back.

I just don't see the use case which makes this a more than a fix for one user. File servers are supposed to be running, not restarted all the time. Otherwise they're generally useless.

comment:3 Changed 3 years ago by yiyihu

It's not frequent failure, It's just kind of protection for occasional network failure. Maybe hardware(NIC reset on guest), maybe software(samba upgrade) etc.

I met this twice, Once is NIC reset for the machine I use under load, Another time is I upgrade my samba server for new version. That's how I know vbox can't handle this kind of error.

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use