Re: 10g Clusterware (Oracle CRS) documentation ?
- From: mhthomas <qnxodba@xxxxxxxxx>
- To: Christo Kutrovsky <kutrovsky.oracle@xxxxxxxxx>
- Date: Thu, 28 Apr 2005 01:15:55 -0400
Hi,
On 4/7/05, Christo Kutrovsky <kutrovsky.oracle@xxxxxxxxx> wrote:
> Hello,
>=20
> I've never heard about the "H3/H1 timeout" in the documentation ? What
> are those ?
>=20
>=20
1) Sorry, I just saw your reply. The H3/H1 timeout are when
/etc/inittab CRS processes do not start properly, they are identified
as H1, H2, H3 . The errors are cryptic and the H's can get out of sync
if you read some of the Notes and Patch Notes. Its challenging to get
RAC restarted without reboot after these errors occur.
2) I recently came across some info on crs_profile from www.oracle.com
under technology/tech/linux/vmware/cookbook/stage4<whatever> regarding
setup RAC on LINUX on VMWare. I do not use the OCFS so I modified my
install accordingly (no ocfs rpm, etc).
I simplified the tuning steps w.r.t. 10g SE/RAC/ASM/RAW (no ocfs)
under VMWare 5 Build 13124. There were many failures of VIP and NIC
/until/ I performed the steps in the following script (basically from
the cookbook). However, be careful because you do not need many other
cookbook tips.
[root@rac71pub root]# cat mike_extract
# root alias:
# alias crshomebin=3D'/u01/app/oracle/product/10.1.0/crs_1/bin'
#
echo ...
echo "... 1) crshomebin"
crshome
echo ...
echo "... 2) Extract the resource from OCR to text:"
echo ...
echo "... ./crs_stat -p ora.rac71pub.vip > /tmp/ora.rac71pub.vip.cap"
./crs_stat -p ora.rac71pub.vip > /tmp/ora.rac71pub.vip.cap
echo ...
echo "... 3) Update modify the .cap file. Change CHECK_INTERVAL to 900s and=
"
echo "... SCRIPT_TIMEOUT to 600s (default is 60 for both):"
echo ...
echo "... ./crs_profile -update ora.rac71pub.vip -dir /tmp -o ci=3D900,st=
=3D600"
echo ...
echo "... 4) Register the updated resource with OCR (notice -u, update alre=
ady"
echo "... registered resource"
echo ...
echo "... ./crs_register ora.rac71pub.vip -dir /tmp -u"
echo ...
echo "... 5) CRS startup on reboot? Increase CSS misscount (def 60s)"
echo "... With this value a disk IO can take up to 180*.75 =3D 160s"
echo ...
echo "... crshomebin"
echo "... ./crsctl set css misscount 180"
echo ...
echo "... 6) Check current misscount:"
echo ...
echo "... ./crsctl get css misscount"
echo ...=20
echo "... 7) Note:"
echo "... Ref: Note 276782.1"
echo ...
echo "... cd ~"
cd ~
echo ...
Now, my 10g RAC under VMWare is working great, even with the VMWare v5
swapping everything to disk and my guest RAM of 696MB (may increase
back to 1G) on RHEL3U2 LINUX.
3) I am confused by the memory configuration on XP Host with VMWare 5
and I'm not sure how the VM memory preferences work. I edited my
Preferences|Memory|Fit all virtual machine memory into reserved host
RAM, but I see two different results on XPSP1 host. Under Task Manager
under Processes Tab the vmware-vmx.exe Mem Usage =3D 678,172K, but
Performance Tab Commit Charge Total =3D 285,356K. Therefore, it appears
my laptop still has lots of memory available and uses only 283MB total
(?). I do not know why the Process Tab Mem Usage does not sum up to
Commit Charge Total.
Regards,
Mike Thomas
PS: I think there is an internal Oracle Note 276782.1 that describes
crs_profile, etc. If anyone can verify/describe/forward that would be
nice.
--
//www.freelists.org/webpage/oracle-l
Other related posts:
