I can reliably reproduce this panic with a similar stack trace on a newly installed Solaris 10 10/08 system (I know, not OpenSolaris but it appears to be the same problem). I just opened a support case w/ Sun but then discovered what appear to be the specific steps for me to reproduce it.
My setup is a Sol10u6 server, with /export/olddata a ZFS filesystem with sharenfs=root=zeus.mattwilson.local zeus.mattwilson.local is an Ubuntu Linux system. I mount the NFS share with no options, just mount athena:/export/olddata /mnt What I think is causing the problem is that if I copy a file, as root, with owner UID 4294967294 to the Solaris NFS share, using the -a option to GNU cp on the Linux box (which, among other things, preserves the owner), the panic occurs. Other files, with more "reasonable" owners, don't panic the server. In my case I can avoid the problem by fixing the bad owner ID on the file I'm copying, but not sure if this helps with your situation. My stack was: SolarisCAT(vmcore.2/10X)> stack unix:vpanic_common+0x165() unix:0xfffffffffb84d7c2() genunix:0xfffffffffb9f0c63() zfs:zfs_fuid_table_load+0xac() zfs:zfs_fuid_init+0x53() zfs:zfs_fuid_find_by_idx+0x87() zfs:zfs_fuid_map_id+0x47() zfs:zfs_fuid_map_ids+0x42() zfs:zfs_getattr+0xbc() zfs:zfs_shim_getattr+0x15() genunix:fop_getattr+0x25() nfssrv:rfs4_delegated_getattr+0x9() nfssrv:rfs3_setattr+0x19d() nfssrv:common_dispatch+0x5b8() nfssrv:rfs_dispatch+0x21() rpcmod:svc_getreq+0x209() rpcmod:svc_run+0x124() rpcmod:svc_do_run+0x88() nfs:nfssys+0x16a() unix:_sys_sysenter_post_swapgs+0x14b() -- switch to user thread's user stack -- panic string: assertion failed: 0 == dmu_bonus_hold(os, fuid_obj, FTAG, &db), file: ../../common/fs/zfs/zfs_fuid.c, line: 95 On Tue, Sep 9, 2008 at 7:56 AM, Mark Shellenbaum <[EMAIL PROTECTED]> wrote: > David Bartley wrote: >> On Tue, Sep 9, 2008 at 11:43 AM, Mark Shellenbaum >> <[EMAIL PROTECTED]> wrote: >>> David Bartley wrote: >>>> Hello, >>>> >>>> We're repeatedly seeing a kernel panic on our disk server. We've been >>>> unable to determine exactly how to reproduce it, but it seems to occur >>>> fairly frequently (a few times a day). This is happening on both snv91 and >>>> snv96. We've run 'zpool scrub' and this has reported no errors. I can try >>>> to >>>> provide more information if needed. Is there a way to turn on more >>>> logging/debugging? >>>> >>>> -- David >>>> -- >>> Have you been using the CIFS server? You should only be going down that >>> path for Windows created files and its trying to load Windows domain SID >>> table. >> >> No. We have a bunch of linux NFS clients. The machines mount from the >> server using a mixture of NFSv3, NFSv4, sys auth, and krb5 auth. >> > > What is the history of this file system? Was is created prior to snv_77 > and then upgraded? You most likely have a bad uid/gid on one or more files. > > Can you post the dump so I can download it? > > -Mark > _______________________________________________ > zfs-discuss mailing list > zfs-discuss@opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discuss > -- Matthew R. Wilson http://www.mattwilson.org/ _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
