Re: [SSI] /usr/bin/cmount
Brought to you by:
brucewalker,
rogertsang
From: Aneesh K. K.V <ane...@di...> - 2002-07-30 03:17:03
|
Hi, I was using the --bind option of mount for Context Dependent Symbolic Links (CDSL). That is /var/log on node1 is actually /var/log-1 and /var/run on node1 is /var/run-1. and similarly on node2 /var/run-2. That means when doing discover_mount() I should not pick /var/log and /var/run because they are different for different nodes. If not done this way many of the use space application may need modifications. Are there any other way i can look at CDSL. On Tru64 it appears as /var/log -> cluster/members/{memb}/tmp and when namei convert the name to inode it will expand {memb} to member0, member1 etc -aneesh On Mon, 2002-07-29 at 22:50, David B. Zafman wrote: > > Actually, I was planning on making --bind mounts SSI (every node mounts > it too) at some point. We have a real problem if we need some to NOT be > mounted on other nodes. Why do you need mounts which can only be seen > on one node? This isn't SSI. I think this is also harder to do then > just distributing everything. I was going to add a new file system type > for unsupported file systems. For example, if a floppy is mounted with > msdos file system only one node really mounts it, but all other nodes > maintain a place holder file system at the mount-point which returns > errors when you try to traverse. This way all nodes have a consistent > tree even when a file system is only accessible from one node. I was > hoping to keep this kind of thing to a minimum, because it breaks > migration for processes that use files in these file systems. > > Aneesh Kumar K.V wrote: > > Hi, > > > > With /usr/bin/cmount how we should handle the CDSL ( directories > > mounted with --bind ). I don't want then to be mounted on any other > > nodes. > > > > Is the best way is to check the bind flags inside the > > ssisys_discover_mounts and don't do a mount on those entries. ? > > > > > > -aneesh > > > > > > > > > > > > > > ------------------------------------------------------- > > This sf.net email is sponsored by:ThinkGeek > > Welcome to geek heaven. > > http://thinkgeek.com/sf > > _______________________________________________ > > ssic-linux-devel mailing list > > ssi...@li... > > https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel > > > > > -- > David B. Zafman | Hewlett-Packard Company > Linux Kernel Developer | Open SSI Clustering Project > mailto:dav...@hp... | http://www.hp.com > "Thus spake the master programmer: When you have learned to snatch > the error code from the trap frame, it will be time for you to leave." > > > > ------------------------------------------------------- > This sf.net email is sponsored by: Dice - The leading online job board > for high-tech professionals. Search and apply for tech jobs today! > http://seeker.dice.com/seeker.epl?rel_code=31 > _______________________________________________ > ssic-linux-devel mailing list > ssi...@li... > https://lists.sourceforge.net/lists/listinfo/ssic-linux-devel |