And good luck if you need any sort of status update, when the SR and bug
are assigned to wholly separate groups within Oracle. I suspect that black
holes let more information escape...
On Fri, Dec 11, 2015 at 4:04 PM, Ls Cheng <exriscer@xxxxxxxxx> wrote:
Agreed, I have many stories to tell about the "one vendor, one support" :-)
Just to list a few:
1. I have a customer who upgraded a couple of years ago from Solaris
10 to 11, Oracle Database 10.2.0.5 RAC to 11.2.0.4 RAC. As we know a couple
of years ago Solaris 11 was like a Beta Version released to the Market, the
customer suffered node eviction everyday, Database Support points to O.S
and hardware. O.S points to RDBMS team. Hardware pointed to OS and
Database. After 1 month pointing each other finally they found a hidden who
knows what Solaris Kernel parameter which improved the issue so all the
fingers pointed to O.S. After 5 months applying every single SRU released
every month they finally stabilized the platform.
2. I have another customer who migrated Windows to Oracle Linux 6.4 in
MAA architecture. A month after installation the database started to see
block corruptions with no reason (it was not even production yet), they
opened 4 SR against Database Support and 2 SR against Oracle Linux Support
and after 7 months pointing to each other someone in database support
finally found the problem, it turns out he upgraded the kernel in his test
system and could not longer reproduce the issue anymore. So it was OL 6.4
default UEK kernel which caused the corruptions
3. I have another customer who runs everything in Oracle Stack. Sun
T5, Oracle Database, Solaris 11. A month ago they had severe performance
problem due to lost blocks in the private interconnect. The opened SR to
RDBMS, Solaris and hardware. Solaris team and hardware team didnt even
bother look the SR, they directly said it was RAC problem even after the
customer showed evidence with OS statistics that there were problem in the
network layer in the OS (network packets needed reassembling all the time).
Finally the customer started to unplug cables and it turned out that it was
a hardware problem, the network card was sort of broken (it loses packages
but not 100% dead).
Lesson learned by these customers? That each Oracle support team works
like a independent company, there is no difference deal with only Oracle or
with Oracle, Red Hat, EMC. The support team just dont talk between them.
