Solved: Where is the partner interface options getting pulled from?

mdvillanueva · ‎2012-09-14

Hi

After running HA config checker, we got this message.

"Interface sm-storage91 (sm-storage91) reffered by filer2 for failover is missing on filer1"

I look interface configuration under the Partner Interface (see attached diagram), and I believe it is supposed to choose sm-storage-91 (10.x.x.11) instead of sm-storage91. There is a missing 'dash'

Where are the options in the drop-down list getting pulled from? from the host file? How do I receive the error? Do I change here first in the interface configuration and then change the /etc/rc file also?

Weird thing is, we were able to perform takeover/giveback before without any issue.

Thanks,

Maico

scottgelb · ‎2012-09-14

Thank you… a few too many replies and got caught in the weeds J This clarifies it a lot and the IPs are matching what you said .11 on node1 and .21 on node2. I would manually fix the partner statement in node2 /etc/rc, then rerun the ifconfig command to set the correct partner so takeover/giveback works properly. That will fix the error and also ha-config check should report it is ok…and nice that it caught there was an issue which is only in the partner parameter on node2.

View solution in original post

scottgelb · ‎2012-09-14

The download page says that it looks for a partner parameter in the rc file "Checks /etc/rc on each storage controller to see that all interfaces have a failover set" ... although I wouldn't trust that alone since it doesn't mean someone ran an ifconfig to remove that parameter after boot (or maybe rc was edited after)... I check the rc file for partner and also ifconfig to make sure it is set as per listed in rc. Does your rc file have a partner parameter on sm-storage-91?

mdvillanueva · ‎2012-09-14

Hi Scott,

Here is the rc entry in filer2

ifconfig sm-storage-91 `hostname`-sm-storage-91 netmask 255.255.255.0 partner sm-storage91 mtusize 9000 trusted -wins

Here is the host entry in filer 2 (you can see that it is different from what shows above. Also in the diagram I added in my other post, the second option in the dropdown list is the oen showing the right name of the interface)

10.x.x.11 filer1-sm-storage-91

Here is the rc entry in filer 1

ifconfig sm-storage-91 `hostname`-sm-storage-91 netmask 255.255.255.0 partner sm-storage-91 mtusize 9000 trusted -wins

Maico

scottgelb · ‎2012-09-14

This looks ok…as long as “ifconfig sm-storage-91” shows the partner interface is set. I wonder if the ha config has an issue when “partner” is not at the end of the line.

mdvillanueva · ‎2012-09-14

Yes it is set in ifconfig but on the gui, when I look at the drop-down list it seems like the second one is the correct one to choose.

However, I am not sure if I should change on the ifconfig or on rc file or both.

Maico

scottgelb · ‎2012-09-14

On reboot or takeover/giveback whatever is in rc/hosts will match the ifconfig so looks like something (probably the gui changed it) modified the ifconfig after... I would change on both to match... although make sure a maintenance window or no outage depending on any syntax issues (and bmc/rlm/sp or console access in case the network does drop)

mdvillanueva · ‎2012-09-14

Right now, rc and ifconfig are the same. Host is different.

If I edit only in rc and then perform a reboot, will it change it in ifconfig?

Maico

scottgelb · ‎2012-09-14

ahh...good catch on the missing dash...definitely fix that.

joker_morgan · ‎2012-09-14

If I understand what you are asking.. and if this a 7mode machine (Assume it is since you are talking takeover/giveback)and you are hitting it with System Manger I would start there as an issue. When reading rc files with the gui we have had nothing but trouble... more accurately said writing and reading has been an issue. Anyway, if you look at the files directly and compare that to an ifconfig -a at the console do things match that way? If they match are they correct or are they both wrong?

mdvillanueva · ‎2012-09-14

Hi joker,

Here is the output in filer2 in regards to the interface. They don’t match to what I have in the host file.

sm-storage-91: flags=0x6b4c863<UP,BROADCAST,RUNNING,MULTICAST,TCPCKSUM,NOWINS> mtu 9000

inet 10.x.x.21 netmask 0xffffff00 broadcast 10.x.x.255

partner sm-storage91 (not in use)

ether xx:xx:xx:xx:25:f6 (Enabled interface groups

Maico

scottgelb · ‎2012-09-14

Was a manual ifconfig or system manager used to change the networking? Like joker_morgan said there have been some issues with System Manager and networking issues when making changes.

joker_morgan · ‎2012-09-14

we have also had issue where the name gets too long. for example we would name our grp FL1-IFG1 and then when the vlan would get added it would end up FL1-IFG1-(vlan Number) and this caused issues being read correctly. We ended up with much shorter names. This was back on 8.0.1 but thought I would throw it out there.

mdvillanueva · ‎2012-09-14

This was configured by a third-party. I believe they configured this manually via script, therefore subject to typo.

Maico

scottgelb · ‎2012-09-14

I would reconcile the current setting and the hosts file value… then change the one that is incorrect and test out a takeover/giveback and make sure still the same IP. A takeover/giveback will re-run the rc file though and will come up with the value in the hosts file which may not be the right one if not set to that now.

mdvillanueva · ‎2012-09-14

If you go back to the diagram I included in the first post,

In the dropdown list, all the other options has the IP in parethesis, except what was chosen.

In filer1, the partner interface that was chosen has an IP address in parenthesis also.

We will perform a takeover/giveback tonight. Should I change both in rc and ifconfig?

Maico

joker_morgan · ‎2012-09-14

forgive me if I am being slow here... but I am trying to make sure I follow you Scott (I am sure you are right by the way, just wanting to learn a little). The RC file on Filer1 has the missing dash so add it there and then (in a very controlled way in case of issue) source that line or perform so take over giveback to allow it all to be reread. I am not understanding what would be getting changed in the hosts file.

mdvillanueva · ‎2012-09-14

Joker,

no, filer2 is the one missing a dash in the rc. the ifconfig also is missing a dash on the “sm-storage91”

scottgelb · ‎2012-09-14

The rc looks ok...using hostname substitution it should be using the .11 address "`hostname`-sm-storage-91"... but it is set to .21 and I don't see .21 set anywhere in rc or hosts.

mdvillanueva · ‎2012-09-14

let me clarify a few things about the IPs

filer 1 is .11 and filer2 is .21

in filer 1,

here is the entry in rc. looks like the hostname does not need to be included.

ifconfig sm-storage-91 `hostname`-sm-storage-91 netmask 255.255.255.0 partner sm-storage-91 mtusize 9000 trusted –wins

Here is the entry in hosts

10.x.x.11 filer1-sm-storage-91

10.x.x.21 filer2-sm-storage-91

Here is what in filer 2

here is the entry in rc.

ifconfig sm-storage-91 `hostname`-sm-storage-91 netmask 255.255.255.0 partner sm-storage91 mtusize 9000 trusted –wins

Here is the entry in hosts

10.x.x.11 filer1-sm-storage-91

10.x.x.21 filer2-sm-storage-91

scottgelb · ‎2012-09-14

Thank you… a few too many replies and got caught in the weeds J This clarifies it a lot and the IPs are matching what you said .11 on node1 and .21 on node2. I would manually fix the partner statement in node2 /etc/rc, then rerun the ifconfig command to set the correct partner so takeover/giveback works properly. That will fix the error and also ha-config check should report it is ok…and nice that it caught there was an issue which is only in the partner parameter on node2.

mdvillanueva · ‎2012-09-14

But what is weird though, I have been using wiregauge to run haconfigcheck and it never catches this and it has been like this forever and we have done failover before without issue.

But the Netapp tech caught this.

Maico