Hi Jeremy. What version of autofs are you using with RHEL6? ------- Yemi On Oct 17, 2014, at 1:47 PM, Jeremy Filizetti wrote: We use it autofs frequently without too many issues. The initial mounts in Lustre 2.x are far slower then 1.8 and I haven't really dug into why because most of my clients are still 1.8. Not sure what you mean by "scalable" for automounting. There was a race with accessing an OSC before it was setup that you can find details in: https://jira.hpdd.intel.com/browse/LU-631 You need to make sure you escape the second colon (after the NID). Add your lustre map to /etc/auto.master and create the mounts in your map similar to: site1 -fstype=lustre,user_xattr :site1-mgs-ib0@o2ib0\:/site1 site2 -fstype=lustre,user_xattr :site2-mgs-ib0@o2ib0\:/site2 Jeremy On Fri, Oct 17, 2014 at 2:48 PM, Adesanya, Adeyemi <yemi@slac.stanford.edu<mailto:yemi@slac.stanford.edu>> wrote: Hi. I would like to revisit the possibility of automounting Lustre. Is there a reliable, scalable automount option for Lustre RHEL6 clients? The man page for 'autofs' states that type 'lustre' is unsupported........ ------- Yemi _______________________________________________ HPDD-discuss mailing list HPDD-discuss@lists.01.org<mailto:HPDD-discuss@lists.01.org> https://lists.01.org/mailman/listinfo/hpdd-discuss

It would be great if others confirmed RHEL6 automounting is working for them too. ------ Yemi On Oct 21, 2014, at 2:35 PM, Adesanya, Adeyemi <yemi@slac.stanford.edu<mailto:yemi@slac.stanford.edu>> wrote: Just ran an automount test (over ethernet) on a RHEL6 client and it looks good. Using the autofs-5.0.5-109 Red Hat package running on 2.6.32-431.29.2.el6.x86_64 kernel. Lustre client and server version are both 2.5.1. Created this entry in the direct mount map: /autofs_lustre_test -fstype=lustre ppa-mds01:/pfs Here's the debug output from the automount daemon when the mount is triggered. Looks straightforward: ----------------------------------------------------------------------------- attempting to mount entry /autofs_lustre_test lookup_mount: lookup(file): looking up /autofs_lustre_test lookup_mount: lookup(file): /autofs_lustre_test -> -fstype=lustre ppa-mds01:/pfs parse_mount: parse(sun): expanded entry: -fstype=lustre ppa-mds01:/pfs parse_mount: parse(sun): gathered options: fstype=lustre parse_mount: parse(sun): dequote("ppa-mds01:/pfs") -> ppa-mds01:/pfs parse_mount: parse(sun): core of entry: options=fstype=lustre, loc=ppa-mds01:/pfs sun_mount: parse(sun): mounting root /autofs_lustre_test, mountpoint /autofs_lustre_test, what ppa-mds01:/pfs, fstype lustre, options do_mount: ppa-mds01:/pfs /autofs_lustre_test type lustre options using module generic mount_mount: mount(generic): calling mkdir_path /autofs_lustre_test mount_mount: mount(generic): calling mount -t lustre ppa-mds01:/pfs /autofs_lustre_test ----------------------------------------------------------------------------- Ran some test transfers OK. Here's the debug output when the mount path eventually expires: ----------------------------------------------------------------------------- expiring path /autofs_lustre_test umount_multi: path /autofs_lustre_test incl 1 unmounting dir = /autofs_lustre_test expired /autofs_lustre_test ----------------------------------------------------------------------------- I also checked the trigger would remount after unmounting. ------ Yemi On Oct 20, 2014, at 7:12 PM, Adesanya, Adeyemi wrote: Hi Jeremy. What version of autofs are you using with RHEL6? ------- Yemi On Oct 17, 2014, at 1:47 PM, Jeremy Filizetti wrote: We use it autofs frequently without too many issues. The initial mounts in Lustre 2.x are far slower then 1.8 and I haven't really dug into why because most of my clients are still 1.8. Not sure what you mean by "scalable" for automounting. There was a race with accessing an OSC before it was setup that you can find details in: https://jira.hpdd.intel.com/browse/LU-631 You need to make sure you escape the second colon (after the NID). Add your lustre map to /etc/auto.master and create the mounts in your map similar to: site1 -fstype=lustre,user_xattr :site1-mgs-ib0@o2ib0\:/site1 site2 -fstype=lustre,user_xattr :site2-mgs-ib0@o2ib0\:/site2 Jeremy On Fri, Oct 17, 2014 at 2:48 PM, Adesanya, Adeyemi <yemi@slac.stanford.edu<mailto:yemi@slac.stanford.edu>> wrote: Hi. I would like to revisit the possibility of automounting Lustre. Is there a reliable, scalable automount option for Lustre RHEL6 clients? The man page for 'autofs' states that type 'lustre' is unsupported........ ------- Yemi _______________________________________________ HPDD-discuss mailing list HPDD-discuss@lists.01.org<mailto:HPDD-discuss@lists.01.org> https://lists.01.org/mailman/listinfo/hpdd-discuss _______________________________________________ HPDD-discuss mailing list HPDD-discuss@lists.01.org<mailto:HPDD-discuss@lists.01.org> https://lists.01.org/mailman/listinfo/hpdd-discuss _______________________________________________ HPDD-discuss mailing list HPDD-discuss@lists.01.org<mailto:HPDD-discuss@lists.01.org> https://lists.01.org/mailman/listinfo/hpdd-discuss