Description

 Solaris 11.2.8.4.0 and 11.2.9.5.0 fail to install on Sun Server X4-8 systems due to the ramdisk(7D) becoming full during the installation, resulting in a system panic in fmd(1M).

Occurrence

This issue can occur when attempting to install, or upgrade to, the following releases on Sun Server X4-8 systems:

x86 Platform

• Solaris 11.2.8.4.0 through 11.2.9.5.0

Note: This issue does not impact Solaris 10 or SPARC systems or any x86 systems other than Sun Server X4-8 systems.

Symptoms

While attempting to install these releases from the ISO image on the X4-8, the system will enter the maintenance mode and the installation will fail, because the fmd crashes and the ramdisk is filled up by the core dump.

When "pkg update" is used to upgrade the system to these releases, the upgrade will succeed. But after the reboot, the OS will enter the maintenance mode because of the fmd crash, with the following dump (example):

      libtopo.so.1`topo_node_name+6(0)
      x86pi.so`x86pi_enum_entity_node+0x20(aecd5b0, 0, efa05654, 0)
      x86pi.so`x86pi_gen_bboard+0x33a(aecd5b0, ad962c0, 2, 0, 3, 0)
      x86pi.so`x86pi_enum_gentopo+0x3ec(aecd5b0, b05c190)
      x86pi.so`x86pi_enum_start+0xd6(aecd5b0, ee741a20, efa04890, efa0834a)
      x86pi.so`x86pi_enum+0x73(aecd5b0, b05c190, c4a4168, 0, 0, 0)
      libtopo.so.1`topo_mod_enumerate+0xd1(aecd5b0, b05c190, a9ffde8, c4a4168, 0)
      libtopo.so.1`enum_run+0xb5(a3f49b0, a3ea8a8, ee741b38, f41a6fb2)
      libtopo.so.1`topo_xml_range_process+0x10f(a3f49b0, 86a5838, a3ea8a8,
      ee741b58)
      libtopo.so.1`tf_rdata_new+0x105(a3f49b0, a9ffe00, 86a5838, b05c190)
      libtopo.so.1`topo_xml_walk+0x2a9(a3f49b0, a9ffe00, 86a5d88, b05c190)
      libtopo.so.1`topo_xml_enum+0x50(a3f49b0, a9ffe00, b05c190, f4193371)
      libtopo.so.1`topo_file_load+0x2d1(a3f49b0, b05c190, b35c718, c4a47a0, 0, 2c)
      libtopo.so.1`topo_tree_enum+0xfa(8607b40, a4faf30, baec083, f41a2aa2)
      libtopo.so.1`topo_tree_enum_all+0x32(8607b40, 204, 0, 0)
      libtopo.so.1`topo_snap_create+0x22e(8607b40, ee741eac, 0, f41a02a8)
      libtopo.so.1`topo_snap_hold_flag+0xc2(8607b40, 0, ee741eac, 1, a2ae0d8,
      a2ae0e0)
      fmd_topo_update+0x9a(80a5dd8, 0, a7fb59b5, d1, a7d0b78, a7d0b88)
      fmd_dr_event+0x216(a7d0b48, ee741f04)
      sysev_legacy+0x93(a7d0b48)
      libsysevent.so.1`subscriber_event_handler+0x9b(baac808, bd942d0, ee741fc8,
      808d54f)
      fmd_thread_start+0x72(bd942d0, f43ef000, ee741fe8, f43ace17)
      libc_hwcap1.so.1`_thrp_setup+0x99(f41c4a40)
      libc_hwcap1.so.1`_lwp_start(f41c4a40, 0, 0, 0, 0, 0)

Workaround

There is no workaround for this issue.

This issue is addressed in the following release:

x86 Platform

• Solaris 11.2.10.5.0 or later

History

21-Apr-2015: Document released; status is Workaround
17-May-2015: Solaris update, issue is Resolved

This regression was caused by the change delivered by bug 18658941.

Questions regarding this document should be addressed to
sunalertpublication_us_grp@oracle.com and copy the
submitter/responsible Engineer listed below.

Internal Contributor/Submitter: james.north@oracle.com
Internal Eng Responsible Engineer: james.north@oracle.com
Oracle Knowledge Analyst: david.mariotto@oracle.com
Internal Eng Business Unit Group: Systems RPE
Internal Associated SRs:  3-10566416487, 3-10565713131

References

This solution has no attachment