cmquerycl and cmcheckconf problems hpux 11v3 and SG 11.18

[jose da silva]

cmquerycl now works but when I do the cmcheckconf it fails to create the binary file. I have attached a txt file with the cmquerycl and cmcheckconf outs and also the cluster ascii file with also the sysinfo out put of the node I am trying to create the cluster.

[Asif Sharif]

Hi Silva,

Same thread,

http://forums11.itrc.hp.com/service/forums/questionanswer.do?threadId=1078448

Regards,
Asif Sharif

[jose da silva]

Yes Asif it is the same. The nodes can now talk to each other but I still have problems as per the attached txt file.

[likid0]

Looks like you have a comunication problem:

uranus lan0 172.26.5.54 successfully received from neptune lan3 172.26.5.58
but neptune lan3 172.26.5.58 did not receive from uranus lan0 172.26.5.54.

uranus lan0 can't talk to neptune lan3 card, you should linkloop the nics and confirm they are working both ways.

for example in uranus:

linkloop -i 0 LAN3MACOFNEPTUNE

[jose da silva]

Hi Asif

Sorry the previous thread is :-
http://forums.itrc.hp.com/service/forums/questionanswer.do?threadId=1325775

[Stephen Doud · Accepted solution]

1) The error messages of the form,
Error reading device /dev/dsk/c9t2d1s3 0x8

... are due to disk devices not longer visible. ioscan will show such disks as NO_HW.

rm or rmsf those special device files if they no longer point to a real disk.

2) The message:
"Volume group /dev/szwvg is configured differently on node uranus than on node neptune"

... is usually due to a mismatch between /etc/lvmtab files where the VG is concerned.
Use 'strings /etc/lvmtab' on both nodes to compare the device special files. More than likely, the node where you imported them will show a complete list, whereas the node where the VG was created will only show one path to each disk. That is the node where you should vgextend the alternate links into the volume group.

3) The message:

"Non-uniform connections detected,
neptune lan0 172.26.5.59 successfully received from uranus lan3 172.26.5.55
but uranus lan3 172.26.5.55 did not receive from neptune lan0 172.26.5.59.
This could be due to heavy network traffic, or heavy load or routing configuration on neptune."

may be the result of several factors.
a. /etc/cmcluster/cmclconfig is invalid for this cluster. If the file was copied from an old cluster, remove it.
b. mismatched speed/duplex of a NIC to switch port or other NICs on the network
Use 'lanadmin -x ' to inspect the speed/duplex of a NIC
c. duplicate (non-unique MAC addr. or IP addr. on a NIC.
Use 'lanscan' to inspect
d. APA (autoport aggregation) installed but not configured
e. Standby NIC is configured with an IP (seen when a cmclconfig already exists)
f. Cluster nodes have large routing tables
Check: netstat -rn
h. Serviceguard does not support more than 4 APA ports per trunk.
I believe uranus has 5<---

[jose da silva]

Hi Stephen

Thanks for your input. I just need some clarification on some of the important points you have mentioned.

2. What happened with the volume groups is that the customer created new volume groups on node neptune before we started the vgexport and vgimports. This caused a problem because the minor numbers were already in use by other vg's and when importing the map files of the clustered vg's I had to create new minor numbers in order to be able to import the vg's.
Stephen will this impact the cluster config if yes how do I go about to resolve this issue . Please give me details on how to solve this if possible.

h. "APA" , I tried to use the each NIC port for example the customer wanted 2 heartbeats, 2 stationary " one for each package" and 2 for standby. So what I did I gave each port an ip address. IF APA is the way to go forward then I will need your assistance and guide on how to go about doing this for the cluster.

f. " Routing tables" YEs you are correct about the routing tables being too large.How can I simplify and correct this issue?

Stephen if you need more info in order so that you can assist and guide me please let me know. Unfortunately I do not have remote login to the servers so I will have to do the changes tomorrow morning. The problem is the time difference, I do not know how many hours difference between ZA and the USA.It would be great if we could keep in touch as I do the changes to the nodes as per your request.Please send me the necessary steps on how to solve the above issues.

Once again THANKS and await your responce.

[jose da silva]

Hi All

I had to reconfigure the netconf file and ensure that the heratbeat ip's have different subnets. Applied cmwuerycl vi the cluster.ascii file did the necessary changes, test the ascii file with cmquerycl and all is fine now. Service Guard heartbeat ip's need to be in different subnets.

Thank you all for your replies for it directed me in the correct path to find a solution.