On Mon, 31 May 2004 00:15:22 +0200, Michael Heiming <[EMAIL PROTECTED]> wrote:
Not in real time with rsync, I suppose it should be fine running the cron job every 15-30 minutes. There's not much changing data on the LTSP server, so even once a hour would be enough.
That really depends on what you are using LTSP to. In my scenario, people are always saving files in their homes, so if doesn't get sync'ed instantaneously, they might loose some data.
There are chances of losing data, but it's very low and a maximum of 30 minutes would be fine and then chances are big it's still on the other server, not really lost, unless the user starts working again on the same file on the other server, thus overwriting the original one a soon as the other node comes up again and rsync starts again.
The goal is to present the user with his environment on the other system. Case the server hw breaks, the tc will probably only display the grey Window with the large X.
No. It might lock entirely, because NFS is mounted with locking. I'm exploring nolock option for NFS, so the user can press ENTER and the system reboots automatically.
Sounds good, albeit users are "experienced" enough to simply switch off/on their tc.;)
Also, since you are connecting your X to an external XDMCP (using -query) it won't even show the grey screen, because tc can't connect. This has been discussed on [EMAIL PROTECTED]
Yep, probably, albeit it doesn't matter (for users) why their tc doesn't work anymore.
Now, user switches off his tc and on again, in about 20 sec. he'll be able to login again (on the second LTSP server) and find his environment.
Yup, but not with rsync, unless you set it to cron, say, 5 minutes, or less. That causes overload. Albeit, 'unison' should be a better choice if you want to go with rsync..
Yep, forgot about unison, thx for the reminder, I'll look about that in addition.
Users don't change their environment (KDE) every five minutes, so there's a >99% chance that in case of fail over, the contents of ~/.kde* will be exactly the same on both nodes. Perhaps, it was not clear what I meant with environment. That's enough ha for my purposes.
Sure GFS or alike would offer better ha, but then I'm limited to run RH ES 3.0 or SuSE SLES 8 (http://www.sistina.com/products_gfs_distros.html), albeit I was thinking about SuSE 9.1, the box is currently running SuSE 8.0. It would need in addition 2 hba per system and need setup multi-pathing.
[..]
There are dozens of distributed filesystems. Some are looking at shared storage devices, some don't. I'm testing CODA, as it seems to do replication, which what I/we want. I've tried OpenAFS, but it's very difficult to install.
Another one could be ENBD (http://www.it.uc3m.es/~ptb/nbd/), haven't used it with LTSP, but the author is a friend of mine, so I presume it working quite well.;)
Thx for the URL you mailed!
Michael Heiming
-------------------------------------------------------
This SF.Net email is sponsored by: Oracle 10g
Get certified on the hottest thing ever to hit the market... Oracle 10g. Take an Oracle 10g class now, and we'll give you the exam FREE.
http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click
_____________________________________________________________________
Ltsp-discuss mailing list. To un-subscribe, or change prefs, goto:
https://lists.sourceforge.net/lists/listinfo/ltsp-discuss
For additional LTSP help, try #ltsp channel on irc.freenode.net
