Patch-ID# 112563-17 Keywords: core admin patch pmfadm support Synopsis: Sun Cluster 3.0: Core/Sys Admin Patch Date: Aug/31/2004 Install Requirements: Install in Single User Mode Reconfigure immediately after patch is installed Solaris Release: 9 SunOS Release: 5.9 Unbundled Product: Sun Cluster Unbundled Release: 3.0 Xref: This patch available for Solaris 8 as patch 110648 Topic: Sun Cluster 3.0: Core/Sys Admin Patch Relevant Architectures: sparc BugId's fixed with this patch: 4363867 4378553 4432402 4522796 4626022 4632588 4634940 4635297 4638586 4644220 4646367 4648767 4653166 4653788 4654399 4657833 4660479 4661799 4665886 4674790 4679834 4680614 4681353 4682326 4683596 4685548 4690047 4690111 4690244 4699854 4700317 4701036 4703481 4704805 4707189 4716500 4719082 4721905 4722715 4727068 4727848 4729785 4734424 4738923 4740024 4748889 4756973 4762272 4766016 4770855 4778869 4781064 4783135 4786012 4792570 4793401 4793429 4797776 4800080 4800870 4803684 4805717 4808610 4808748 4809944 4810418 4811232 4819742 4831151 4835819 4836380 4837987 4844784 4851276 4854173 4855930 4862321 4863012 4863063 4864001 4865182 4868043 4869835 4879228 4880376 4924098 4926752 4931902 4940649 4951767 4964849 4967566 4978602 5024937 Changes incorporated in this version: Patches accumulated and obsoleted by this patch: 116592-01 Patches which conflict with this patch: Patches required with this patch: Obsoleted by: Files included with this patch: /etc/cluster/clpl/SUNW.adapter.bge.clpl /etc/cluster/clpl/SUNW.generic.clpl /etc/init.d/initrgm /etc/rc0.d/K05initrgm /etc/rc1.d/K05initrgm /etc/rc2.d/K05initrgm /etc/rc3.d/S29initrgm /etc/rcS.d/K05initrgm /kernel/drv/did /kernel/drv/sparcv9/did /kernel/fs/pxfs /kernel/fs/sparcv9/pxfs /kernel/misc/cl_comm /kernel/misc/cl_dcs /kernel/misc/cl_dlpitrans /kernel/misc/cl_net /kernel/misc/sparcv9/cl_comm /kernel/misc/sparcv9/cl_dcs /kernel/misc/sparcv9/cl_dlpitrans /kernel/misc/sparcv9/cl_net /kernel/strmod/cldlpihb /kernel/strmod/sparcv9/cldlpihb /usr/cluster/bin/pmfadm /usr/cluster/bin/pnmd /usr/cluster/bin/pnmset /usr/cluster/bin/scdidadm /usr/cluster/bin/scgdevs /usr/cluster/bin/scswitch /usr/cluster/bin/scvxinstall /usr/cluster/lib/dcs/scconf_vxvm.so.1 /usr/cluster/lib/dcs/scvxvmlg /usr/cluster/lib/libclcomm.so.1 /usr/cluster/lib/libdid.so.1 /usr/cluster/lib/libdid_ccr.so.1 /usr/cluster/lib/libdid_ccrfile.so.1 /usr/cluster/lib/libdid_compat.so.1 /usr/cluster/lib/libdsdev.so.1 /usr/cluster/lib/libpmf.so.1 /usr/cluster/lib/librgm.so.1 /usr/cluster/lib/libscconf.so.1 /usr/cluster/lib/libscswitch.so.1 /usr/cluster/lib/rgm/rt/hastorageplus/hastorageplus_postnet_stop /usr/cluster/lib/rgm/rt/hastorageplus/hastorageplus_prenet_start /usr/cluster/lib/rgm/rt/rgoffload/rgofl_probe /usr/cluster/lib/rgm/rtsupport/5.9 /usr/cluster/lib/sc/chkinfr /usr/cluster/lib/sc/cmm_ctl /usr/cluster/lib/sc/replctl /usr/cluster/lib/sc/reserve /usr/cluster/lib/sc/rgmd /usr/cluster/lib/sc/rpc.fed /usr/cluster/lib/sc/rpc.pmfd /usr/cluster/lib/sc/run_reserve /usr/cluster/lib/sc/scds_pmf_action_script /usr/cluster/lib/sc/scsi /usr/cluster/lib/sc/sparcv9/rpc.pmfd /usr/cluster/lib/scadmin/defaults /usr/cluster/lib/scadmin/lib/scinstall_interactive /usr/cluster/lib/sparcv9/libdsdev.so.1 /usr/cluster/lib/sparcv9/libpmf.so.1 /usr/sbin/didadm Problem Description: Remove the fixes for bugs 4990364 and 5057283. (from 112563-16) 4729785 BAD TRAP in module pxfs due to a NULL pointer dereference (attr_tokencache) 4926752 race between extending write and mmap I/O 4964849 ufs : **WARNING: ignoring hole in write 4967566 rcp to pxfs can cause file corruption. (removed) 4990364 global vxfs filesystem is landing into I/O error state after failover (removed) 5057283 panic: BAD TRAP in module "SUNW,UltraSPARC" due to a NULL pointer dereference (from 112563-15) 5024937 reserve with fence_node option dumps core (from 112563-14) 4808748 assertion failed: cur_data_token & PXFS_WRITE_TOKEN 4811232 scconf -rq causes cluster panic in some situations 4924098 mmap of pxfs files can hang on Solaris 9 update 2 and later on 32-bit kernels 4940649 Cache growing without bound 4951767 device fencing code does not check for local_only 4978602 SC3.0: Can't add quorum device due to flaw in quorum_check_transition (from 112563-13) This revision created to accumulate and obsolete patch 116592-01 (from 112563-12) 4378553 RGM does not enforce an RG's Nodelist property on a cluster reboot. 4781064 scdidadm -R panic in cleanup_subpath function 4783135 Shared disks fail to be recognized as DID devices. 4786012 RM cannot handle addition of multiple RMA's simultaneously 4805717 SC 3.0 Oracle error during scsetup 4835819 Unable to switch Device Group : Some Volumes are busy 4837987 Node 5 can not start with Invalid adapter_id for adapter - sci0 message 4844784 SC 3.0 deadlock at sc_per_tick_processing() while shutdown 4851276 scconf -a -D takes long time with large number of LUN system. 4855930 infrastructure corrupted after removal of a cluster node was being interrupted 4862321 init 6 can bring cluster node to the ok prompt instead of rebooting 4863012 mtime changed on global vxfs after file restored 4863063 bad mutex panic after pxvfs::empty_inactive_list destroy a pxnode 4864001 deadlock in pxfs module during growfs operation. 4868043 DCS removes and writes elements to ccr serially 4869835 run_reserve fails if DID namespace is not in sync 4879228 pxfs fobj_ii::lock_deleted function leads to panic by accessing a destroyed obje 4880376 SC3.0 open() returns ENOENT while vxdctl enable/vxassist is running 4931902 DID driver: didprop_op panics with dev_t value of DDI_DEV_T_ANY (from 112563-11) 4626022 fsck output should not be printed on the console. 4719082 Must syslog return values from operations that can fail such as mount 4653788 with Failover_enabled=FALSE, failover occurs on START failure during restart 4679834 global mount failure for VxFS 3.5 4748889 vxfs globally mounted filesystem with vxfs3.5 failed with switchovers 4770855 vxfs3.5 filesystem mounted globally get into I/O error state during switchover 4703481 incorrect file attributes reported after using setext (chgsize) 4699854 scvxinstall will not work with VxVM 3.5 as the license command changed 4707189 scvxinstall tries to install VRTSvmdev, even with VxVM 3.5 4716500 scconf/scsetup fails to register VxVM dgs w/enclosure names as device groups 4854173 SC 3.0 fencing code needs support for VxVM 3.5 4793429 single-node resourcegroup blocks "scswitch -S -h ..." 4740024 `scswitch -S` leads to livecache rg landing in stop_failed state 4778869 remote RG dependencies have negative impact when restarting RG 4819742 Inappropriate memory usage in error handling code of libscswitch. 4836380 scshutdown command coredump in sc3.1U1_09 4797776 simultaneous fsync and memory mapped I/O may result in deadlock 4800080 Cannot see files from NFS clients on global NFS exported filesystems 4831151 SC3.0: 2nd node panic'd with "pure virtual function called"; 1st node at OBP 4865182 Remove FSYNC specification for direct I/O based read/writes 4810418 Remote direct io write performance for Oracle RAC needs to be improved (from 112563-10) 4661799 scstat status message changed after data service ported to DSDL 4685548 GDS continually restarts failing applications 4792570 SC tcp-transport does not work with bge devices 4793401 gds binary core dumps 4800870 NAFO failover does not work with ce interfaces 4803684 Private interconnect discovery does not work well with bge devices 4808610 scds_pmf_action_script dumps core for RT's that don't declare failover_enabled 4809944 reserve hangs when using enclosure-based names (from 112563-09) This patch revision was created to correct a patch packaging problem. The deliverable objects are exactly the same as in the previous revision. (from 112563-08) 4683596 did reconfiguration takes too long on configs with many disks 4722715 SC3.0 takes longer to switchover than SC2.2 on a large scale environment. (from 112563-07) 4762272 scswitch -R causes panic if resource goes stop_failed and failover_mode != HARD 4756973 rgmd uses idl object after failed idl call in scha control giveover: causes segv 4766016 SC 3.0 does not support bge ethernet interface as cluster interconnect (from 112563-06) This patch revision was created to backout the fixes for bugs 4678077 and 4620185. (from 112563-05) 4635297 fuser doesn't show any processes running on a global raw device (removed) 4678077 Unable to trace, gcore, pstack a process under cluster control (removed) 4620185 rpc.pmfd: race condition between pmfadm -k and forks by a monitored process 4701036 cluster 3.0 scalable service mcnet module cannot heavy UDP load 4704805 denial of service attack by sending fragmented packets to shared address 4721905 topology manager ignores error returned from wrsm driver 4734424 Add ce to the list of known public net devices in SUNWscr posinstall 4738923 downloading 40MB file causes node panic "getapage: NO read lock even for a page" 4700317 "lockfs -f" is hanging on a pxfs (from 112563-04) This patch revision was created to correct a patch packaging problem. There are no functional differences between this revision and the previous one. (from 112563-03) 4681353 A5000 disk drives are reserved after SC3.0 installation 4727068 qualified dataservice listed as "non-qualified" through scinstall 4727848 HACI private commands need to be shipped with SUNWscu in sc30 patch (from 112563-02) 4363867 Private IP interfaces have wrong broadcast address 4632588 kernel stack overflow when mounting vxfs fs global in EMC config 4634940 pxvfs::vget() should check for UNMOUNTED 4682326 Running tc_directio dTET test case results in a hang 4680614 node hangs while running S75MOUNTGFSYS in sc3.1/build 33/SVM 4690047 ioctl failure during sendfilev() tests panic the cluster node 4690111 deadlock in DSDL while restarting the resource (through PMF) 4690244 Failfast: Aborting because "rgmd" died 30 seconds ago (from 112563-01) 4432402 Cannot access Photon disk through /dev/did path after replacement w/o reboot 4674790 Disk replacement with no reboot doesn't handle swapped disks 4522796 Support NLM_SHARE with PXFS 4638586 scconf does not recover from VXVM minor-number conflicts with patch 110648-12 4644220 rgoffload is not offloading failover rg when 1 node down in a 2-node cluster 4646367 rpc.fed limits filedescriptors to 1024. 4648767 rw_exit: lock not held, lp=3002fab4fd8 wwwh=0 thread=3002f5aba60 4665886 suncluster 3.0 panics in pxfs with recursive rw_enter 4653166 readlink returns different results on UFS vs PXFS 4654399 scrgadm -r -j is hung on cluster taurus 4657833 when auto_start_on_new_cluster is false, failover is not happening. 4660479 nlm clustering hooks do not recover non-pxfs locks (from 116592-01) 4880376 SC3.0 open() returns ENOENT while vxdctl enable/vxassist is running -------------------------------- There are three possible procedures for installing patches on Sun Cluster 3.0. Additional instructions for this patch are specified below in the "Special Install Instructions" section. Refer to the chapter entitled "Patching Sun Cluster Software and Firmware" in the Sun Cluster 3.0 System Administration Guide for a description of the different installation processes and instructions on how to install Sun Cluster 3.0 patches. For Solaris 9 release, refer to the man pages for instructions on Patch Installation Instructions: -------------------------------- There are three possible procedures for installing patches on Sun Cluster 3.0. Additional instructions for this patch are specified below in the "Special Install Instructions" section. Refer to the chapter entitled "Patching Sun Cluster Software and Firmware" in the Sun Cluster 3.0 System Administration Guide for a description of the different installation processes and instructions on how to install Sun Cluster 3.0 patches. For Solaris 9 release, refer to the man pages for instructions on using 'patchadd' and 'patchrm' scripts provided with Solaris. Special Install Instructions: ----------------------------- Install this patch using the patch install procedure for a "Rebooting Patch (node)" as described in the "Patching Sun Cluster Software and Firmware" chapter of the "Sun Cluster 3.0 System Administration Guide". Note 1: In order to run supported network adapters which use the ce(7D) driver under Public Network Monitor (PNM) NAFO groups, add the following line to the /etc/system file on all the cluster nodes and reboot them. set ce:ce_reclaim_pending=1 Bug 4766016 ----------- To get the complete fix for bug 4766016 you must also install patch 114189-01 (or newer). Bug 4803684 ----------- This patch enables Sun Cluster to work with the bge ethernet device. When installing Sun Cluster on a cluster with one or more bge devices on the private interconnect, the following special procedure needs to be followed. 1. Install Sun Cluster packages from the CD using 'scinstall -i'. This will install the packages but will not perform the Sun Cluster configuration steps. 2. Install this patch. 3. Perform steps 1 and 2 on all new cluster nodes before proceeding to the next step. 4. Run scinstall again, this time without any options. When you are asked whether you want to use autodiscovery for private interconnect configuration, choose no. Autodiscovery does not work well for bge devices. Please use the manual option to enter the private interconnect configuration. README -- Last modified date: Tuesday, August 31, 2004