|
|
19571359
|
Pillar- Zpool is Suspended on Various I/Os
Failover or Failback Tests
On an Oracle Virtual Networking chassis running two servers, the
I/O failover or failback does not work on the Pillar storage but
works on the ZFS, Netapp, and 3PAR storages. This bug is not an
Oracle Virtual Networking bug.
Workaround: None.
|
20443969
|
PVI Multicast Join or Attach
Fails
When a PVI is created on a Linux host first and then on an Oracle
Solaris host, the host fails to be multicast-capable. This is
because the default MTU of the Hermon driver is set to 2000.
Workaround: You must first
create the PVI on the Oracle Solaris host.
|
17490439
|
Failed to Detect LUN 0 When MPXIO Is
Enabled From the Host
After the vHBA is created, Oracle Solaris 11.3 hosts fail to
detect LUN 0 when LUN 0 is added to a vHBA. This is a SCSAv3 layer
bug.
Workaround: LUN 0, which is the
controller LUN, should always be mapped to the vHBA manually.
|
20419596
|
No TCP Traffic to Oracle VM Server for
SPARC on a Checksum-Enabled Jumbo PVI
When checksum-enabled vNICs or PVI vNICs are pushed to Oracle VM
Server for SPARC that has connectivity with Oracle Solaris 11.3
hosts, there will be no TCP traffic if the vNIC or vSwitch in the
Oracle VM Server for SPARC has a non-zero checksum value configured.
This is not an Oracle Virtual Networking issue.
The same vNIC works fine in the primary domain. By default,
checksum is not enabled on a vNIC and hence works fine.
Workaround: None.
|
19516042
|
Zpool Goes Into Unavailable Mode After
Host Reboot
During the boot process, Zpool goes into a faulty state when all
the vHBAs are yet to be online. Zpools come online after FMA becomes
operative. However the faulty state affects the zones. This is not
an Oracle Virtual Networking issue.
Workaround: Use the following
command to bring the Zpool online:
zpool clear
zpool-name.
|
20461377
|
xsvnicMust Know the
Device State After Install
There might be an obstruction to xsvnic's DR
when the port is open for querying the link state. This is because
the link state details are updated only when the
mc_start function is called.
Workaround: None.
|
20461341
|
Package Uninstallation Fails to Clean
up
xstn
Module From the
ib.conf
File
Because of the unavailability of the post
remove option in IPS, during installation, the Oracle
Virtual Networking service start method tries to
update the ib-conf file by using the
cfgadm_ib command. This issue does not have
any impact as xstn module and other Oracle
Virtual Networking modules are removed after uninstallation.
Workaround: None.
|
19054744
|
TCP Responses on an Aggregation Interface
Can Sometimes Be Prohibited
Some tools and utilities force a vNIC or PVI vNIC into promiscuous
mode. On an Oracle Solaris interface that is aggregating (for
example, trunking) PVI vNICs that have checksumming enabled, a
problem can prevent a response to TCP packets sent over the
interface. This problem occurs because the tool forces the vNIC or
PVI vNIC into promiscuous mode.
Workaround: None.
|
18963387
|
Secondary I/O Domain Attached to Multipath
Group Disk Does Not Complete Reboot
With two I/O domains attached to two virtual disks as a multipath
group (one domain per disk), the secondary I/O domain can get
“stuck” during the bootup sequence when an Oracle Solaris 11.3 host
is rebooted. When the secondary I/O domain is “stuck,” it does not
complete its reboot and never comes back online.
Workaround: Edit the
/etc/system file for all I/O domains
connected to the Oracle Solaris host to include the
following:
firceload: drv/PX
|
18877148
|
Problem When Setting an Oracle Virtual
Networking Server Profile Down
An internal logic-handling error in the Oracle Solaris SCSAv3 code
can cause a kernel panic in Oracle Solaris 11.3 hosts when you
attempt to set a server profile to down state.
Workaround: None.
|
18086995
|
The
ipadm delete-ip
Operation Pushes vNIC to Up/Down State in
Chassis
For vNICs or PVI vNICs that are added from the Oracle Fabric
Interconnect, if you issue the ipadm delete-ip
command on an Oracle Solaris 11.3 host to remove the network address
for a vNIC, a problem sets the vNIC to up/down state on the Oracle
Fabric Interconnect. This problem affects standard vNICs and PVI
vNICs.
Workaround: You can work around
this problem by assigning the IP address from the host.
|
18046241
|
vNIC Host Managed IP Address for vNIC Is
Not Displayed on Chassis
If you configure a vNIC, then attempt to assign the vNIC’s IP
address as a host-managed address, the IP address is visible on the
host. However, a problem prevents the address from being displayed
on the Oracle Fabric Interconnect when you issue the show
vnic command.
Workaround: None.
|
18046181
|
Do Not Allow Duplicate PVIs With the Same
Net ID
A problem allows duplicate PVUs to be created with the same Net
ID. This is incorrect, as each PVI should have its own unique Net
ID.
Workaround: None.
|
18046171
|
No Traffic between Access vNIC and Host
Managed on Trunk vNIC
An access-mode PVI connected to one Oracle Solaris 11.3 host and a
trunk PVI connected to a different Oracle Solaris 11.3 host cannot
pass traffic through the same VLAN when both PVIs are connected to
the same cloud.
Workaround: None.
|
18018632
|
PVI Cloud vNICs Remained in
Up/Initializing
After Server Profile
Disconnect/Reconnect
When two servers have 16 vNICs and 16 PVIs configured across them
(8 vNICs and 8 PVIs on each server), and traffic is occurring on all
16 vNICs, if the server profile is disconnected and reconnected, the
PVIs might not come back up. Instead, a problem causes them to
remain in up/initializing state. This problem is seen only on the
PVIs. The standard vNICs come back online correctly after the server
profile is connected.
Workaround: If you encounter
this problem, you can work around the PVI problem by setting the
PVIs down, then up.
|
17940487
|
PVI Loss in ICMP Packets When Host Comes
Up After Reboot
When a host is rebooted and the PVI is coming up on the host and
the Oracle Fabric Interconnect, ICMP packet loss occurs during
pings, but only for a short time. ICMP packet loss goes from 100% on
the first ping session to 0% packet loss by the third ping
session.
Workaround: None.
|
17927174
|
Data-Link Manager Failed to Add a vNIC
Device
On an Oracle Solaris server that has a full 16 vNICs (8 standard
vNICs and 8 PVI vNICs), a problem can prevent all vNICs from being
displayed by data-link management processes. When the problem
occurs, all vNICs are shown by other processes, but data-link
management misses one vNIC. For example, prtconf
-D correctly shows all 16 vNICs, but dladm
show-phys shows only 15 vNICs.
Workaround: None.
|
17877733
|
PVI vNIC and Clouds Statistics Are Not
Incremented
In an HA Oracle Fabric Interconnect deployment (where multiple
servers are redundantly connected to two Oracle Fabric
Interconnects), statistics for PVI vNICs and PVI clouds are not
incrementing. This problem causes the following commands to show
zero:
Workaround: None.
|
17487287
|
System Crashes When vHBAs and vNICs
Removed at Once
A problem can cause Oracle Solaris 11.3 hosts to crash in some
situations while vHBAs are being deleted. This problem occurs if you
delete multiple vHBAs and immediately attempt to delete vNICs while
the vHBA deletion process is still ongoing. Also, this problem can
occur when you issue the format command on the
host while the vHBA deletion process is still ongoing.
Workaround:
-
Delete vNICs before deleting vHBAs, or if you need to
delete vHBA first, wait until all vHBAs have been completely
deleted before starting to delete vNICs.
-
Do not run the format command on the
host while vHBAs are being deleted.
|
17370928
|
LUN Masking Is Not Working to Show a Mask
of LUNs Behind a Target
In this release, LUN masking is not supported from the Oracle
Fabric Interconnect. Do not use the LUN mask feature from the Oracle
Fabric Interconnect because it will not properly mask, and all LUNs
will be visible.
Workaround: Configure any LUN
masking from the storage target.
|
17337836
|
Stale Entries getting Created in
the
/dev/
Directory
When LUNs and targets are dynamically added or deleted, entries
are written to the /dev directory. However, a
problem prevents the proper cleanup of these entries, and as a
result, stale entries can accumulate in /dev.
When enough entries have accumulated, issuing the
format command hangs.
Workaround: Periodically,
manually clean up the entries in /dev. Issue
the following command:
devfsadm -Cv
|
17254225
|
LUN 0 Must Always Be Mapped to the vHBA
for SCSAv3
Per SCSAv3, LUN 0 (zero) is required and should be mapped to vHBAs
for correct reporting. However, the host drivers do not check or
enforce this requirement, so it is possible that LUN 0 is not mapped
to the vHBAs. If LUN 0 is not mapped to vHBAs, inconsistencies can
occur on those vHBAs.
Workaround: Make sure that LUN
0 is mapped to vHBAs.
|
17234437
|
LUN Not Visible
informat
Command From NetApp
Storage
A problem prevents LUNs on NetApp storage from being visible when
issuing the format command on an Oracle Solaris 11.3 host.
Workaround: You can work around
this problem by adding lines to the
scsi_vhci.conf file on the Oracle Solaris
host. Follow this procedure:
-
Log in to the Oracle Solaris 11.3 host, and using
vi, emacs, geddit, ot any standard
text editor, open the
/kernel/drv/scs_vhci.conf file for
editing.
-
Find the scsi-vhci-failover-override =
statement and add the following NetApp line:
"NETAPP LUN", "f_sym"
|
165306626
|
Change the LUN Mask Not Detaching the Old
LUN Mask's LUN
In a LUN mask assigned to an Oracle Solaris host, the original
LUNs remain masked even if the configuration of the LUN mask has
been changed. If you edit the LUN mask, the changes do not take
effect. For example, if you create a LUN mask with two LUNs, then
remove those LUNs and add two more, the LUN mask will incorrectly
contain four LUNs (the two newly added LUNs, plus the two LUNs from
the original LUN mask). This problem occurs even after rescanning
the vHBA.
Because LUN masking is not assigned dynamically, you must set the
vHBA down, then up.
Workaround: Do the following
when you make any changes to a LUN mask that is already
created:
-
Set the vHBA down:
set vhba
name.server-profile-name
down
-
Set the vHBA up again:
set vhba
name.server-profile-name
up
|
16338290
|
Error Seen in
dmesgon vNIC Creation and
Deletion
When adding or deleting a vNIC interface, spurious messages are
displayed on the Oracle Solaris host. Be aware that messages are
displayed, but they are not always errors. You need to scan the
messages to determine if an actual error exists. For example, the
following messages are actual errors:
@ Jan 25 10:17:29 sparc1-prb nwamd[756]: [ID 588122 daemon.error]
1:@ nwamd_set_unset_link_properties: dladm_set_linkprop(mtu)
failed for net25:@ operation not supported
@ Jan 25 10:17:29 sparc1-prb nwamd[756]: [ID 387169 daemon.error]
1:@ nwamd_unconfigure_interface: disable failed for net25:
Operation failed
Workaround: No workaround
currently exists for this problem.
|