Patch-ID# 120500-28

6268548 clexecd does not handle many parallel requests
 
(from 120500-27)
 
6907541 one node of cluster hangs while shutting down after executing `scshutdown`
6987225 panic in orb invocation code due to a freed context
 
(from 120500-26)
 
6824685 Sun Cluster 3.1 postremove and preremove scripts don't handle alternate root for Live Upgrade
6898395 when applying latest KU 141444-09 clusters hang on reboot as they exhaust autopush structures
6933172 bad trap panic in cl_flk_state_transition_notify
6934273 PATH environment variable limit of 1024 prevents GDS resource start
6970105 read-ahead logic in PxFS can cause ufs_rdrw_data() to consume all memory on x86 with Sun Cluster 3.1
6984835 data integrity issues can occur in PxFS on Sun Cluster 3.1 when writing data with an offset beyond the EOF
 
(from 120500-25)
 
6829261 Solaris Cluster needs to support new TCP/IP hooks
6886841 PxFS returns file attributes on Solaris 9 NFS client when the file has been removed via rm(1)
 
(from 120500-24)
 
4502242 rcm_daemon reports errors during pkgadd of VRTSvxvm
4520248 rcm_daemon reports errors with SDS
6500349 "scdidadm -r" dumps core with a SEGV
 
(from 120500-23)
 
6800955 scshutdown hangs due to the inconsistent values of rwlocks_granted
 
(from 120500-22)
 
        This revision fixes patch construction issues.
 
(from 120500-21)
 
6406472 pxreg::putpage waiting in page_lock_es - both nodes hung
6485394 deadlock panic during lock manager recovery
6634592 nss_cluster mods to match new nsswitch API in S10U4
6672126 private interconnect using IXGE - commented ixge_taskq_disable=1 should to be set in /etc/system
6736639 new NULL frange_lock panics caused by CR 6226001
6749960 run_reserve's use of dgconv does not scale well
 
(from 120500-20)
 
6487117 RFE: Sun Cluster should automatically request for intr mode RX processing for private interconnects
6617974 customer needs to be able to set the mtu for jumbo frames on Virtual public interfaces in Sun Cluster
 
(from 120500-19)
 
6454600 _owner != threadid() assertion failure hit in mutex lock
6569237 devfsadmd is dumping core very frequently during functional tests
6582506 SC3.2/S10U3/AMD - BAD TRAP during cluster initialization in 8-node cluster
6599044 Huron (UltraSPARC T1) platforms are facing panics when InfiniBand HCA are installed or configured as interconnects
6700365 upon scshutdown, scswitch -S hangs in wait_for_unfreeze
6713370 node panic during SCATE startup
6722684 closeconn_timer_manager destructor can prevent realtime callouts from executing
 
(from 120500-18)
 
6226001 BAD TRAP: cluster: alignment error
6298053 cannot safely link applications against libstlport4 on Sun Cluster
6497226 repl_pxfs_server::become_primary sees NULL mnt_vp and panics
6531464 cl_apid core dumping due to not-fully initialized sc_event class
6595309 panic due to LOFS+PxFS
6605376 DID driver debug stuff needs some cleanup
6606946 mdboot() hang waiting for cmm_callback_impl::wait_for_stopped()
6622212 cl_apid is core dumping
6644815 did driver does not release bp_mapin buffers for some I/O
6660245 double free in scswitch
6664671 closeconn_timer_manager destructor may deadlock with callout threads
6688814 3.2 scinstall -u update -R /mnt modifies files on /
 
(from 120500-17)
 
6244400 clprivnet handling of UDP fragments is suboptimal
6579880 SC private transport version discovery algorithm suboptimal
 
(from 120500-16)
 
6471022 network/multipath service remains in maintenance mode until IPMP groups are created, state cleared
6514047 LD_PRELOAD of alternate malloc() causes SC admin commands to crash
6521832 scshutdown: umount2() of /global/.devices/node@X required
6583942 DID driver must use dev_open always with OTYP_LYR
 
(from 120500-15)
 
6370086 race condition in hafoip binaries leading to cluster node reboot
6389571 ipmp failure of unrelated group leads to shared address resource group failover
6477926 SC3.1U4/S10/SPARC - nodes panic with quorum issue during reboot with extended SAN solution
6543864 PxFS deadlock "panic: Deadlock: cycle in blocking chain"
6548416 scgdevs takes long time to finish (>5hrs for ~600lun)
6580371 fix for CR 6507810 incomplete; causes data corruption
 
(from 120500-14)
 
6497075 Sun Cluster does not handle correctly remote USCSICMD ioctls
6507810 data corruption with PxFS
6517625 RGM debug messages not written to syslog after integration of CR 6374957
 
(from 120500-13)
 
6346569 RG stuck in PENDING_OFFLINE; R is in STOPPING_DEPEND
6429808 RG switchover fails due to INACTIVE flag set on interface
6438132 rgmd dumped core while resources were being disabled
6479096 scalable_sticky does not work as designed
6479731 scstat -i reports 'unknown' after integration of 6429808
 
(from 120500-12)
 
6489678 fix for CR 6468041 could remove ipnodes setting for user
 
(from 120500-11)
 
6268054 ntp.conf.cluster errors during install
6346472 autodiscovery needs to recognize ipge on SPARC platforms
6353533 tc_all_reboot scate test failed to shutdown one node
6408500 ipnodes entry in nsswitch.conf not updated during scinstall
6414195 new interface nge not recognized with Sun Cluster scinstall
6414826 add Kirkwood NIC (PCI-X 10GB Ethernet) support for Sun Cluster/Solaris 10 as transport adapters
6427799 nscd causes bad trap due to access to not yet initialized ORB
6435295 cluster process spending a lot of time in rxdoor_bucket::find
6449139 autodiscovery needs to recognize e1000g driver on SPARC systems
6460419 syntax error in scswitch kills rgmd
6468041 scinstall should expand tryagain status to other sources of ipnodes in addition to NIS
6475684 scsnapshop hangs looping in Analyzer.pm at RG_DEP_LOOP
 
(from 120500-10)
 
6252843 RGM fails to enforce RG's Nodelist property on cluster reboot
6285275 panic: incoming reply I/O to buf that has been deleted
6375746 dgconv may fail if local_only flag set to true
6376257 NFS threads blocked in check_wait_lock_cache()
6406877 concurrent append writes and reads to shared file on global UFS lead to garbage data
6431208 /etc/rc2.d/S77scpostconfig.sh should not have "exit 0" statement
6442814 change reconfigure to reboot on README for Sun Cluster core patches
 
(from 120500-09)
 
4986143 scvxvmlg reports non-existent error
6299971 scinstall should add/delete /etc/cluster/nodeid in filelist.ramdisk during install/uninstall
6419670 scinstall -r failed with "error reading boot archive file"
6334051 async_writes/fsync have window for data token corruption
6236392 pxfs_fi tests tc_fi_bulkio_pages(4,8) failed
6392984 sync_filesystem should not do VN_RELE while holding hash lock
6339206 when discarding attributes, write_back_attr() should clear both access_rights and attr.va_mask
6361530 unable to offline ERROR_STOP_FAILED rg when rg_system=true
6374957 RGM dying when GDS dataservice registers with wls%wls
6401312 *OOM*: memory leak in PxFS
6402020 sccheck makes use of both private and public networks without need
6425083 Logical Host resource not coming online on node 2 when all private interconnects pulled on node 1
 
(from 120500-08)
 
        This revision addresses patch construction issues.
 
(from 120500-07)
 
4480601 dirty pages associated with PxFS mem_async objects can consume all of memory
6234508 improve failure detection when cluster node panics
6392507 offline device groups cause panic during scversions
6376023 '|' character in EMC device ID confuses didadm CCR parsing
6259193 SunMC's daemon scsymon_srv daemon doesn't start by default on Solaris 10; have to start manually
6203133 scinstall does not copy all Cacao security files
6299971 scinstall should add/delete /etc/cluster/nodeid in filelist.ramdisk during install/uninstall
6361979 cladm called by ifconfig causes bad trap panic due to orb not yet initialized
6376995 PxFS sequential write performance with largefiles (>2Gb) copies very slow
6379547 memory leak in cl_eventlogd
 
(from 120500-06)
 
6276503 IPv6 interfaces plumbed will cause scinstall to fail with scrconf: RPC: Remote system error
6322576 wrong netmask (ffffff80) when using 172.16.x.x as public net with S10/SC3.1/u4
 
(from 120500-05)
 
4680500 need support for quotas when using local filesystem
5103486 clprivnet_udreq does not set mblk size correctly in some cases
6262161 unreferenced file on init 6 - needs "smart" pointer to close rgm_vp_version
6199229 enabling heartbeat/interrupt messages via syslog in default bits
6264024 "scswitch -S" fails with "Global service <> associated with path <> found to be in maintenance state
6312828 cluster panics with 'rgmd died' panic when ld_preload set and scstat or scha_resource_get is used
6326325 cluster node panicked with traptype 0x31 in module cl_comm (in new PxFS code)
6346221 ls not showing all files in directory on cluster filesystem
 
(from 120500-04)
 
6263077 assertion failed: attr.va_mask
6301459 nodes hang in four-node clustered pairs load test with faults running HA-Oracle, NFS
6272796 nodes hard hang at boot
6302037 assertion failed: write_token.contains(node_id) at pxfs_v1/server/file_impl_v1.cc, line: 744
6314787 panic[cpu1]/thread=3000335a2e0: HA: hxdoor 11.428 does not exist on secondary
6318319 amd64: surviving node panicked as soon as one node rebooted by faultclient
6320186 excessive dnlc purging causes global filesystem performance regression
 
(from 120500-03)
 
        NOTE: This revision accumulates feature point patch 120551-03 and also
              delivers the following bugfixes:
 
6252931 during quorum test, BAD TRAP: type=e (Page fault (#pf)) rp=f1f2ead4 addr=0 occurred in module "unix"
6257112 public network down leads to hafoip_stop timing out and to node reboot
6264616 check_node_health should check to see if it has enough freemem to continue reconfiguration
6281287 cl_ccrad core dumps leaving SPM unusable for customer
6306113 cluster nodes attempt to use IPv6 enabled transport adapters to access hosts on public network
 
(from 120500-02)
 
6312633 ftruncate fills up file with junk values
 
(from 120500-01)
 
6312445 opening global vxvm device in maintenance state can cause panic (in device_server_v1::find_fobj)
 
(from 120551-03)
 
6215921 assertion failed: env.exception() == NULL, file: cl/pxfs/mount/mount_server_impl.cc, line: 4240
6262494 generic data service does not probe the Network_resources_used/Port_list
6278520 scalable service not working with bge transport on Solaris 10
6302466 packages SUNWscr and SUNWscu add both the same root crontab entry to call newcleventlog
6313806 didadm core dumped on first bootup on second node in two-node cluster
6314983 U4 patch 120550-02 upgrade renders node useless, /usr/cluster/lib/librgm.so.1: symbol __1cDrgmISHUTDOWN
 
(from 120551-02)
 
5067092 assertion HA: hxdoor 11.166 does not exist on secondary
5105711 scshutdown failing when RG has rg_system set to true even if RG offline
6293845 scha_*(1HA) APIs should not generate events since they can potentially cause event storms
6290248 rgmd dumped core while rs stop failed flag cleared after disabling all resources in RG
6285745 scswitch -s -k leaves the resource in 'Stopping' state forever
6227414 sci_configure_adapter_instance: Failed to add sci adapter
6303524 didadm dumps core on second node if the internal RAID is configured on v40
 
(from 120551-01)
 
5068618 node panic during scshutdown: Cluster lost operational quorum
6232265 scshutdown results in panic "assertion failed: owner == lo, file: cl/repl/rmm/rmm.cc, line: 217"
6252769 scshutdown and Oracle RAC
5075946 rpc.pmfd crashes due to writing outside of its allocated memory
6174741 assertfail in os::condvar_t::broadcast
6209791 change hasp to bring online global devices if in scalable RG
6214580 scvxinstall is using wrong device paths when used with VxVM 4.0
6223498 panic in cl_net module if 2 scalable resources use same IP address but different port numbers
6230417 NFS-PxFS write fails "Returning EIO as uio_iovcnt is %d" or "panic: uio_clnt_cmn was too fragmented"
6246125 panic when did_adminioctl released kernel buffer to wrong cache during scdidadm -R operation
6259545 quorum code should "or" __amd64 to existing __i386 ifdef to support amd64 platform
6300732 assertion failed: trans_ctxp == NULL

--------------------------------
 
There are three (3) possible procedures for installing patches on Sun
Cluster 3.1.  The proper method to use and any 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 3.1 System Administration Guide" for a
description of the different install processes and instructions on how
to install Sun Cluster 3.1 patches.
 
Refer to the man pages for instructions on using 'patchadd' and
'patchrm' commands provided with Solaris, to install and remove patches.

-----------------------------
 
NOTE 1:  Install this patch using either of the following patch install
         procedures found in the "Patching Sun Cluster Software and Firmware"
         chapter of the "Sun Cluster System Administration Guide"
 
         * Rebooting Patch (Node)
         * Rebooting Patch (Cluster and Firmware)
 
NOTE 2:  Regardless of which patch installation method is used, if at any time
         this patch must be removed, first shut down the cluster and reboot
         the nodes into non-cluster, single user mode.  This patch cannot be
         removed from a node while any cluster node is in cluster mode.
 
NOTE 3:  To get the complete fix for BugId 5068618 (node panic during
         scshutdown: Cluster lost operational quorum), please also install
         the following patch:
 
         120497-01 (or greater)  Sun Cluster 3.1 OPS Core Patch for Solaris 10
 
NOTE 4:  On systems which are not utilizing IPv6 networking, it is
         recommended that IPv6 entries in /etc/netconfig follows IPv4
         entries.
 
         For example, the line
 
         tcp        tpi_cots_ord  v     inet    tcp    /dev/tcp        -
 
         Should come before the line
 
         tcp6       tpi_cots_ord  v     inet6   tcp    /dev/tcp6       -
 
NOTE 5:  To get fix for bug 6688814 (3.2 scinstall -u update -R /mnt modifies
         files on /) you cannot apply the patch with 'patchadd -d' due to bug
         6678937 (patchadd needs to support patching package level scripts to
         the system install directory).
 
NOTE 6:  If you use Solaris Volume Manager, please also install patch 120485-05
         or later which addresses the issue in CR 6454600.
 
NOTE 7:  If you use lofs, please comment out
 
         #exclude: lofs
 
         in the /etc/system file.