firstly what file system are you using? i can't see this working with any standard non distributed file system.
i've never seen anything like this -- two pc's essentially using a single disk. but everything makes me think that this would not be a good solution as i can't think of any file system that was developed to be used like this. the problem is one of abstraction; the scsi bus handles putting blocks onto the disk, and that is all. the os filesystem functions organise those blocks into an entity that is file system. a user space protocol and daemons such as NFS can handle multiple accesses and sort them out for handing down the chain. i can't see how this setup can guarantee normal unix semantics ; i.e. a read after write returns the value just written and two writes returns the latest write. if this fails, then so does everything built ontop of it. why you can't see updates i don't know, but i think that would only be a symptom of a larger problem. for what it's worth, i can't see this working without some sort of distributed file system, e.g. nfs or samba. failing that, you would have to write your own file system that implements some sort of session semantics, immutable files or atomic transactions or some other scheme you think of yourself. sounds like fun but probably not the easiest way to solve the problem. i'm happy to be corrected on any of the above points, however. -i > -----Original Message----- > From: Grant Parnell [SMTP:[EMAIL PROTECTED]] > Sent: Thursday, April 18, 2002 10:40 AM > To: [EMAIL PROTECTED] > Subject: [SLUG] Sync of filesystem on multiple SCSI hosts > > I have a client with the following setup: > > __________________ > | | > | Firewire storage | > |__________________| > | | > | | > ________|_ _|________ > | | | | > | Server A | | Server B | > |__________| |__________| > > This is not an NFS mount, the box has multiple SCSI buses. > > Both Server A and Server B mount the same filesystem and my client says > that when files are written by Server A the only way they can be seen on > Server B is to unmount the filesystem and re-mount it. I did suggest using > the sync option when mounting the filesystems and also trying the sync > command but this didn't help. > > Effectively we need to lose the linux filesystem buffers (and yes, all the > efficiency that goes with that) I think but I don't know how to do that. > > Possibly this could be done periodically on each server to allow some > level of filesystem efficiency but I'm guessing this would be more > trouble. > > If this isn't easy I'm going to have to suggest NFS... wonder if you can > do NFS over SCSI? I've heard of TCP/IP over SCSI I think. I do not know if > the Servers can see each other on the SCSI buses, certainly not visible in > /proc/scsi areas so my guess is no. > > -- > ---<GRiP>--- > Grant Parnell - senior consultant > For all your Linux Commercial quality support and consulting needs > Web: http://www.linuxhelp.com.au Email: [EMAIL PROTECTED] > For retail sales see http://www.everythinglinux.com.au > Phone 02 8753 0792 to book service. > > -- > SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/ > More Info: http://lists.slug.org.au/listinfo/slug ********************************************************************** CAUTION: This message may contain confidential information intended only for the use of the addressee named above. If you are not the intended recipient of this message, any use or disclosure of this message is prohibited. If you received this message in error please notify Mail Administrators immediately. You must obtain all necessary intellectual property clearances before doing anything other than displaying this message on your monitor. There is no intellectual property licence. Any views expressed in this message are those of the individual sender and may not necessarily reflect the views of Woolworths Ltd. ********************************************************************** -- SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/ More Info: http://lists.slug.org.au/listinfo/slug
