I have a client running 1.4.11 and i want to mount a filesystem that''s running on 1.6.4.1 servers. On the client I can ping the servers using lctl however a mount fails with an LBUG on the mds. The mount command on the 1.4.11 client is -- mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre The error on the server is -- LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- >lov_connects == 1) failed LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG ll_mdt_24 R running task 0 7369 1 7370 7368 (L-TLB) ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 Call Trace: [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 [<ffffffff800867b0>] default_wake_function+0x0/0xe [<ffffffff8005bc25>] child_rip+0xa/0x11 [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 [<ffffffff8005bc1b>] child_rip+0x0/0x11 I thought 1.4 could mount a 1.6 filesystem. Aaron Knister Associate Systems Analyst Center for Ocean-Land-Atmosphere Studies (301) 595-7000 aaron at iges.org
On Jan 21, 2008 12:59 -0500, Aaron Knister wrote:> I have a client running 1.4.11 and i want to mount a filesystem that''s > running on 1.6.4.1 servers. On the client I can ping the servers using > lctl however a mount fails with an LBUG on the mds. > > The mount command on the 1.4.11 client is -- > > mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre > > The error on the server is -- > > LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- > >lov_connects == 1) failed > LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG > ll_mdt_24 R running task 0 7369 1 7370 7368 > (L-TLB) > ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 > 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 > ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 > Call Trace: > [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff > [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c > [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe > [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 > [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 > [<ffffffff800867b0>] default_wake_function+0x0/0xe > [<ffffffff8005bc25>] child_rip+0xa/0x11 > [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 > [<ffffffff8005bc1b>] child_rip+0x0/0x11 > > I thought 1.4 could mount a 1.6 filesystem.Only if the filesystem was previously upgraded from a 1.4 system. If it was formatted as 1.6 initially there is no "downgrade" compatibility in the configuration logs. Cheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc.
Oh, ok! Thanks! -Aaron On Jan 21, 2008, at 6:37 PM, Andreas Dilger wrote:> On Jan 21, 2008 12:59 -0500, Aaron Knister wrote: >> I have a client running 1.4.11 and i want to mount a filesystem >> that''s >> running on 1.6.4.1 servers. On the client I can ping the servers >> using >> lctl however a mount fails with an LBUG on the mds. >> >> The mount command on the 1.4.11 client is -- >> >> mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre >> >> The error on the server is -- >> >> LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- >>> lov_connects == 1) failed >> LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG >> ll_mdt_24 R running task 0 7369 1 7370 >> 7368 >> (L-TLB) >> ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 >> 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 >> ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 >> Call Trace: >> [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff >> [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c >> [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe >> [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 >> [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 >> [<ffffffff800867b0>] default_wake_function+0x0/0xe >> [<ffffffff8005bc25>] child_rip+0xa/0x11 >> [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 >> [<ffffffff8005bc1b>] child_rip+0x0/0x11 >> >> I thought 1.4 could mount a 1.6 filesystem. > > Only if the filesystem was previously upgraded from a 1.4 system. If > it was formatted as 1.6 initially there is no "downgrade" > compatibility > in the configuration logs. > > Cheers, Andreas > -- > Andreas Dilger > Sr. Staff Engineer, Lustre Group > Sun Microsystems of Canada, Inc. >Aaron Knister Associate Systems Analyst Center for Ocean-Land-Atmosphere Studies (301) 595-7000 aaron at iges.org
How about the interoperability of a 1.6 server with 1.4 routers? I knew clients were possible but I haven''t seen any info on routers. Id like to get the servers up to a somewhat new distro but with our XT3 still on 2.6.9 the routers on ithave to stay at 1.4. brian Aaron Knister wrote:> Oh, ok! Thanks! > > -Aaron > > On Jan 21, 2008, at 6:37 PM, Andreas Dilger wrote: > >> On Jan 21, 2008 12:59 -0500, Aaron Knister wrote: >>> I have a client running 1.4.11 and i want to mount a filesystem >>> that''s >>> running on 1.6.4.1 servers. On the client I can ping the servers >>> using >>> lctl however a mount fails with an LBUG on the mds. >>> >>> The mount command on the 1.4.11 client is -- >>> >>> mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre >>> >>> The error on the server is -- >>> >>> LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- >>>> lov_connects == 1) failed >>> LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG >>> ll_mdt_24 R running task 0 7369 1 7370 >>> 7368 >>> (L-TLB) >>> ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 >>> 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 >>> ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 >>> Call Trace: >>> [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff >>> [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c >>> [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe >>> [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 >>> [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 >>> [<ffffffff800867b0>] default_wake_function+0x0/0xe >>> [<ffffffff8005bc25>] child_rip+0xa/0x11 >>> [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 >>> [<ffffffff8005bc1b>] child_rip+0x0/0x11 >>> >>> I thought 1.4 could mount a 1.6 filesystem. >> Only if the filesystem was previously upgraded from a 1.4 system. If >> it was formatted as 1.6 initially there is no "downgrade" >> compatibility >> in the configuration logs. >> >> Cheers, Andreas >> -- >> Andreas Dilger >> Sr. Staff Engineer, Lustre Group >> Sun Microsystems of Canada, Inc. >> > > Aaron Knister > Associate Systems Analyst > Center for Ocean-Land-Atmosphere Studies > > (301) 595-7000 > aaron at iges.org > > > > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss at lists.lustre.org > http://lists.lustre.org/mailman/listinfo/lustre-discuss
On Jan 22, 2008 13:58 -0500, Brian W. Johanson wrote:> How about the interoperability of a 1.6 server with 1.4 routers? I knew > clients were possible but I haven''t seen any info on routers. > > Id like to get the servers up to a somewhat new distro but with our XT3 > still on 2.6.9 the routers on ithave to stay at 1.4.There was one known issue with routing 1.6 clients through 1.4 routers, but that was resolved (sorry, don''t know bug number right now, but it was a public bug from LLNL so should be findable in bugzilla).> Aaron Knister wrote: > > Oh, ok! Thanks! > > > > -Aaron > > > > On Jan 21, 2008, at 6:37 PM, Andreas Dilger wrote: > > > >> On Jan 21, 2008 12:59 -0500, Aaron Knister wrote: > >>> I have a client running 1.4.11 and i want to mount a filesystem > >>> that''s > >>> running on 1.6.4.1 servers. On the client I can ping the servers > >>> using > >>> lctl however a mount fails with an LBUG on the mds. > >>> > >>> The mount command on the 1.4.11 client is -- > >>> > >>> mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre > >>> > >>> The error on the server is -- > >>> > >>> LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- > >>>> lov_connects == 1) failed > >>> LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG > >>> ll_mdt_24 R running task 0 7369 1 7370 > >>> 7368 > >>> (L-TLB) > >>> ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 > >>> 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 > >>> ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 > >>> Call Trace: > >>> [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff > >>> [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c > >>> [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe > >>> [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 > >>> [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 > >>> [<ffffffff800867b0>] default_wake_function+0x0/0xe > >>> [<ffffffff8005bc25>] child_rip+0xa/0x11 > >>> [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 > >>> [<ffffffff8005bc1b>] child_rip+0x0/0x11 > >>> > >>> I thought 1.4 could mount a 1.6 filesystem. > >> Only if the filesystem was previously upgraded from a 1.4 system. If > >> it was formatted as 1.6 initially there is no "downgrade" > >> compatibility > >> in the configuration logs. > >> > >> Cheers, Andreas > >> -- > >> Andreas Dilger > >> Sr. Staff Engineer, Lustre Group > >> Sun Microsystems of Canada, Inc. > >> > > > > Aaron Knister > > Associate Systems Analyst > > Center for Ocean-Land-Atmosphere Studies > > > > (301) 595-7000 > > aaron at iges.org > > > > > > > > > > _______________________________________________ > > Lustre-discuss mailing list > > Lustre-discuss at lists.lustre.org > > http://lists.lustre.org/mailman/listinfo/lustre-discuss > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss at lists.lustre.org > http://lists.lustre.org/mailman/listinfo/lustre-discussCheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc.
I should have figured LLNL has done this already. Unless we compile a 1.6 liblustre, all clients (mostly catamount but a few service nodes) that use the 1.4 lnet routers will be 1.4 thenselves. Thanks Andreas. brian Andreas Dilger wrote:> On Jan 22, 2008 13:58 -0500, Brian W. Johanson wrote: >> How about the interoperability of a 1.6 server with 1.4 routers? I knew >> clients were possible but I haven''t seen any info on routers. >> >> Id like to get the servers up to a somewhat new distro but with our XT3 >> still on 2.6.9 the routers on ithave to stay at 1.4. > > There was one known issue with routing 1.6 clients through 1.4 routers, > but that was resolved (sorry, don''t know bug number right now, but it > was a public bug from LLNL so should be findable in bugzilla). > >> Aaron Knister wrote: >>> Oh, ok! Thanks! >>> >>> -Aaron >>> >>> On Jan 21, 2008, at 6:37 PM, Andreas Dilger wrote: >>> >>>> On Jan 21, 2008 12:59 -0500, Aaron Knister wrote: >>>>> I have a client running 1.4.11 and i want to mount a filesystem >>>>> that''s >>>>> running on 1.6.4.1 servers. On the client I can ping the servers >>>>> using >>>>> lctl however a mount fails with an LBUG on the mds. >>>>> >>>>> The mount command on the 1.4.11 client is -- >>>>> >>>>> mount -t lustre 192.168.64.70 at o2ib:/data-mdtlov/client /mnt/lustre >>>>> >>>>> The error on the server is -- >>>>> >>>>> LustreError: 7369:0:(lov_obd.c:227:lov_connect()) ASSERTION(lov- >>>>>> lov_connects == 1) failed >>>>> LustreError: 7369:0:(tracefile.c:431:libcfs_assertion_failed()) LBUG >>>>> ll_mdt_24 R running task 0 7369 1 7370 >>>>> 7368 >>>>> (L-TLB) >>>>> ffff81026e92ddd0 ffff8102ba45045c ffff810418438dc0 ffff8102823e0200 >>>>> 0000000000000000 ffff810296765000 0000000000000000 ffff81041c2fd280 >>>>> ffffffff888c029f 0000000000000286 ffffffff8003cd0e ffff810418438140 >>>>> Call Trace: >>>>> [<ffffffff888c029f>] :mds:mds_handle+0xe7f/0x56ff >>>>> [<ffffffff8003cd0e>] lock_timer_base+0x1b/0x3c >>>>> [<ffffffff8001c1a2>] __mod_timer+0xb0/0xbe >>>>> [<ffffffff885dcfb0>] :obdclass:class_conn2export+0x250/0x2d0 >>>>> [<ffffffff8868e277>] :ptlrpc:ptlrpc_main+0x1437/0x1df0 >>>>> [<ffffffff800867b0>] default_wake_function+0x0/0xe >>>>> [<ffffffff8005bc25>] child_rip+0xa/0x11 >>>>> [<ffffffff8868ce40>] :ptlrpc:ptlrpc_main+0x0/0x1df0 >>>>> [<ffffffff8005bc1b>] child_rip+0x0/0x11 >>>>> >>>>> I thought 1.4 could mount a 1.6 filesystem. >>>> Only if the filesystem was previously upgraded from a 1.4 system. If >>>> it was formatted as 1.6 initially there is no "downgrade" >>>> compatibility >>>> in the configuration logs. >>>> >>>> Cheers, Andreas >>>> -- >>>> Andreas Dilger >>>> Sr. Staff Engineer, Lustre Group >>>> Sun Microsystems of Canada, Inc. >>>> >>> Aaron Knister >>> Associate Systems Analyst >>> Center for Ocean-Land-Atmosphere Studies >>> >>> (301) 595-7000 >>> aaron at iges.org >>> >>> >>> >>> >>> _______________________________________________ >>> Lustre-discuss mailing list >>> Lustre-discuss at lists.lustre.org >>> http://lists.lustre.org/mailman/listinfo/lustre-discuss >> _______________________________________________ >> Lustre-discuss mailing list >> Lustre-discuss at lists.lustre.org >> http://lists.lustre.org/mailman/listinfo/lustre-discuss > > Cheers, Andreas > -- > Andreas Dilger > Sr. Staff Engineer, Lustre Group > Sun Microsystems of Canada, Inc. > > _______________________________________________ > Lustre-discuss mailing list > Lustre-discuss at lists.lustre.org > http://lists.lustre.org/mailman/listinfo/lustre-discuss