==> Regarding [autofs] Multimount problem; [EMAIL PROTECTED] adds: david> Hello, david> I have a problem with autofs multimount on Linux. david> On our SGI,SUN,IBM these lines works.
david> auto.master david> /users /etc/auto.users david> auto.users david> * -rw,intr,hard caosrv01:/caosrv01/users/& david> caosrv02:/caosrv02/users/& caosrv03:/caosrv03/users/& This is a replicated server entry not a multimount entry. david> On linux, the log reports no errors when I run autofs and the mount david> works correctly (ex : mount caosrv02:/caosrv02/users/toto /r is OK ) OK, so you can mount the file system by hand. david> but when I want to use autofs ( ex : cd /users/toto ) I have an error david> if toto is on caosrv02 or caosrv03 david> Mar 28 10:16:25 nim_client automount[18049]: >> mount: david> caosrv01:/caosrv01/users/toto failed, reason given by server: Permission david> denied You said you get a failure if autofs selects caosrv02 or caosrv03, yet you show caosrv01 failing. Can you mount from caosrv01 manually? david> Mar 28 10:16:25 nim_client automount[18049]: mount(nfs): nfs: mount david> failure caosrv01:/caosrv01/users/toto on /users/toto david> Mar 28 10:16:25 nim_client automount[18049]: failed to mount /users/toto david> Mar 28 10:16:25 nim_client automount[18049]: umount_multi: no mounts found david> under /users/toto david> Mar 28 10:16:25 nim_client automount[18036]: attempting to mount entry david> /users/toto david> Mar 28 10:16:25 nim_client automount[18051]: >> mount: david> caosrv01:/caosrv01/users/toto failed, reason given by server: Permissiondavid> denied david> Mar 28 10:16:25 nim_client automount[18051]: mount(nfs): nfs: mount david> failure caosrv01:/caosrv01/users/toto on /users/toto david> Mar 28 10:16:25 nim_client automount[18051]: failed to mount /users/toto david> Mar 28 10:16:25 nim_client automount[18051]: umount_multi: no mounts found david> under /users/toto david> Mar 28 10:16:58 nim_client automount[18036]: attempting to mount entry david> /users/toto david> If I try this many times (5, 10, 20 tries ), the filesystem comes up. david> I'm using FC4 with autofs 4.1.4.5. david> I tried on several linux and autofs versions and I always had the same david> thing. david> I try to change my auto.users map syntax like this david> auto.users david> * -rw,intr,hard \ david> caosrv01:/caosrv01/users/& \ david> caosrv02:/caosrv02/users/& \ david> caosrv03:/caosrv03/users/& david> OR david> * -rw,intr,hard caosrv01,caosrv02,caosrv03:/server/users/& david> (on each server, /server is a link with /caosrv0X , the /etc/exports is david> updated and mount is OK on the station. ) I'm having difficulties following your problem report, but I suspect that you are running into a limitation of the code. autofs version 4.1.4 does not handle replicated server entries very gracefully. One of the limitations is that if the first server chosen from the list of replicated servers fails to mount, then autofs will not continue on and try the next server. There have been several discussions on this mailing list dealing with the replicated server code. Jeff Layton provided a patch which overhauled that subsystem; you may want to dig up his patch and see if it works for you. The RHEL variants of autofs have a patch to do the automatic retrying described above. However, that patch was written in a way that would be least invasive to the code. I wouldn't suggest that the patch gets accepted verbatim into the upstream tree. Rather, we should take another look at Jeff L's patches. If you believe that your problem is different from the one I've described, then please visit: http://people.redhat.com/jmoyer. Take a look at the section entitled "Filing bug reports," and write back to this list when you have collected all of the relevant information (you have most of it, already). I hope this was helpful. Thanks, Jeff _______________________________________________ autofs mailing list [email protected] http://linux.kernel.org/mailman/listinfo/autofs
