Oracle VM server for SPARC 2.0

  • Support for Oracle’s SPARC T3-1 server.
  • Support for memory dynamic reconfiguration (DR)
  • Support for the PCI Express (PCIe) direct I/O feature
  • Support for cooperative guest domain migration
  • Support for CPU whole-core allocation and affinity binding
  • Support for universally unique identifiers (UUIDs)
  • Support for LogicalDomains agents
  • Support for the virtinfo command and API
  • Support for the primary domain being configured from an XML file.
  • OracleVMServer for SPARC Physical-to-Virtual (P2V) Conversion Tool enhancements.
  • Support for memory power management (PM).
  • Support for a PMpower cap
  • Support for a PMP-state.
  • Support for virtual disk multipathing.

Resolved Issues

Oracle VM Server for SPARC 2.0 RFEs and Bugs Fixed in Oracle Solaris 10 9/10

  • 6669994  Add a domain service to support OS identification – solaris
  • 6680702  RFE: Want a mechanism for Solaris to know if it is an LDOM and if so what type
  • 6699505  vdc_init_attr_negotiation() should compare msglen to the size of vd_attr_msg_t not vio_ver_msg_t
  • 6708257 DVD-ROM (not OS installation disk) can not mount from guest domain of LDOM 1.0.2
  • 6712763 RFE: Need information about control domain from within a guest domain
  • 6717615 Ambiguous treatment of offline cpus with respect to DR
  • 6720954  add memory dr feature to ldoms
  • 6726533 vdisk failover should handle storage/storage-path failures (a la mpxio)
  • 6741065  Solaris should support cooperative migration
  • 6794962  cpu DR incorrectly rejects some attempts to remove vcpus
  • 6813200 Logical Domains Agents
  • 6832285 Recursive mutex panic when failures from hv_ldc_rx_set_qhead() are not handled properly.
  • 6840358 i_ldc_rxq_drain should not call i_ldc_set_rx_head if the queue is empty
  • 6856230 service domain panics when exporting a file as a single-slice disk
  • 6873151 RFE: unique ID for LDOMs to facillitate asset tracking
  • 6873667 a domain running Sun Cluster should notify the cluster framework when it is migrating
  • 6875401 In elastic mode, a guest at the reboot menu can stall ldm commands
  • 6915529 User level Domain service loopback client registration fails in some scenarios
  • 6924052 iostat within guest domains report 100% busy following a control domain reboot
  • 6929496 deadlock with mdeg_notify_clients and mdeg_unregister hit with DEBUG, causes login hang
  • 6931722 vdisk using failfast or multipath can hang during heavy I/O
  • 6931950 vdisk does not timeout when backend does not exist
  • 6934669 virtinfo missing calls to l10n functions setlocale/textdomain
  • 6935018 S10U9 patch for ldoms agents should depend on libds.so.1
  • 6939070 ldmad problem with largefiles (file > 2GB)
  • 6939707 domain can panic on ‘ldm stop’
  • 6940624 Spurious DS console message: Service ‘agent-system’/client already registered
  • 6945577 LDoms DS fails to register SP ‘pri’ if more than 10 active domains booted.

RFEs and Bugs Fixed for Oracle VM Server for SPARC 2.0 Software

  • 6717661 ldmd does not report MAUs (CPUset) in order.
  • 6783343 LDom Manager needs to parse components portion of PRI for rcid
  • 6852143 set-vcc prevented when system already in del-reconf mode
  • 6887611 ldm ls -o checks are not consistent; some check 2 letters, some check the entire key
  • 6903235 RFE: Zeus should support cooperative migration
  • 6904672 ldmp2v should support root file system on slices other than s0
  • 6905204 “WARNING: ncp1: only one instance (0) allowed” seen on doing P2V migration of a sun4u to N2/VF
  • 6908327 ldmp2v prepare usage message could be improved
  • 6908985 DRM: an enabled policy may not decrease cpu count when it needs to
  • 6909998 DRM: bit of inconsistency in assigning tod-begin and tod-end parameter values
  • 6914643 ldmd core dumps on upgrade from 1.2 patch-02 to 1.3 when inactive domains are in existence.
  • 6916428 ldmp2v: Flash archiving method should support excluding file systems, directories and files
  • 6919211 ldmp2v should be aware of /var/fm
  • 6920852 ldmp2v convert can fail if source system has 0.0.0.0 ip address
  • 6922211 Unnecessary message in ldmd SMF log: warning: vswitch node missing vlan-id property
  • 6923698 /usr/sbin/ldmconfig maximum number of ldoms don’t match documentation
  • 6924550 Dring full, vdc/vds hang while “ldmp2v prepare” a domain.
  • 6926557 RFE: Support new version of mdstore with timestamp and config size
  • 6928250 DRM policy handle mismatch with manual
  • 6929164 ldmp2v(1M) collect phase should not ufsdump(1M) on actively mounted FS.
  • 6932645 PM legacy memory leaks plus clean-up
  • 6933260 Mount vdisks on files with -o nologging
  • 6934296 “ldm migrate-dom” is not correctly interpreted
  • 6936397 DRM: is not reducing vcpus down to ‘min-vcpu’ on systems that are basically idle (%util=0.1)
  • 6938200 code to get PM policy should not retry every second when SNMP support can’t be initialized
  • 6942367 LDoms configuration GUI needs to be updated with Oracle branding
  • 6947814 Domain boot failure if target ldmd dies during a migration
  • 6948564 pmi_get_policy_by_snmp() calls often take over a second to complete
  • 6949238 Guest domain hangs after P2V from Solaris 10 update N to update N
  • 6949242 Migration error for incompatible FW versions is not informative
  • 6950417 when DRM policy times out, vcpu count remains at the same level
  • 6951597 add a resource_id to pcie_device path components
  • 6953092 root disk on EMC Powerpath device confuses ldmp2v
  • 6953460 ldoms seems to be querying HV utilization way to often
  • 6953655 LDoms guest domain MAC address and hostid automatic assignment by ldmd is not very random
  • 6953803 ldmd coredump in ASSERT(snmp_hdlp == NULL)
  • 6957982 CPU Defrag needs to lower its frequency of attempts after repeated failures occur.
  • 6959476 ldmd not properly informing SMF start script to restart the service on certain failures
  • 6961659 rm-vdpcs causes delayed reconfig and later core dumps
  • 6962470 Migration state MD version failures not handled gracefully
  • 6962545 A source ldmd crash during migration causes restart to fail
  • 6966817 ldm(1M) needs an execution profile

With all  other issues and work around, I would not touch Oracle VM server SPARC 2.0

About laotsao 老曹

HopBit GridComputing LLC Rockscluster Gridengine Solaris Zone, Solaris Cluster, OVM SPARC/Ldom Exadata, SPARC SuperCluster
This entry was posted in CMT. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s