Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

Bond Creation fails on 2.5

Pages: 1 2

raf.schandevyl@telenet.be
6 Posts
May 11, 2020, 5:30 pm

Hi,

I'm installing an clean 2.5 version.  From the moment that i create a bond, i lose network during config.  On version 2.4 it works perfect.

Regards

Raf

#1

admin
2,930 Posts
May 11, 2020, 8:11 pm

Hi,

thanks for the info, would you please give more detail:

  1. what kind of bond ?
  2. is this on the management network or backend ?
  3. does it happen on deploy of node 1 or 3 ?
  4. does it happen on the page you create the bond or the last deployment step ?

As i understand we do not actually create the bond until the last page/step in the deployment wizard, i may be wrong, but i believe this is the case to allow for the Prev/Next buttons which is also in version 2.4.

Would it be possible to send us log files, to contact-us @ petasan.org

/opt/petasan/log/Petasan.log
/opt/petasan/config/cluster_info.json
/var/log/syslog (preferably just the end/tail part just after the issue)

Another simplistic thing: can you try from a different browser or same one but empty any cached data. This is probably too optimistic but if it is easy to try for now.

 

Last edited on May 11, 2020, 8:12 pm by admin · #2

raf.schandevyl@telenet.be
6 Posts
May 12, 2020, 5:01 am

Hi,

I have this when trying to create a balance-alb / balance-tlb bond on management interface and backend interface.  This happens on the first node, i never got to configuring the second en third node, as i can't contact the first node anymore.

This happens a the last page of the deployment.  I try to send the logs, but it wil be difficult as i lose contact with the server.

I'll try with another browser today

Regards

Raf

#3

admin
2,930 Posts
May 12, 2020, 9:40 am

would it be possible to bond the backend and not the management interface and see if this proceeds ok. i am sure we test this, but it could be a timing issue.

Last edited on May 12, 2020, 9:43 am by admin · #4

raf.schandevyl@telenet.be
6 Posts
May 12, 2020, 10:15 am

Hi,

Same problem i wizar stops at page 7, and no reply to ping.  I'll try it also without setting jumbo-frame.  What look strange for me is that the wizard gives me 10 nic's while there are only 6 in the server.

 

Regards

Raf

#5

admin
2,930 Posts
May 12, 2020, 10:41 am

Yes jumbo frames could be the issue, as you are connecting on management interface with normal MTU, switching to jumbo frames will drop the connection...but we should dis-allow this , maybe it is a new bug that allows changing jumbo frames on management.

the 10 nic issue is intentional: this is included in a page for creating the entire cluster, not necessarily the node you are currently connected to, before we use to demand all nodes have the same number of interfaces, but now we offer the flexibility of nodes having different setups based on what roles you give them. It probably needs some clarification in the ui, it could be a bit confusing.

#6

raf.schandevyl@telenet.be
6 Posts
May 12, 2020, 10:44 am

I was only setting jumbo frame on backend bond, not on the management bond.  when not setting jumbo frame ik seems to work.

 

Regards

Raf

#7

admin
2,930 Posts
May 12, 2020, 11:23 am

Thanks for the info. we did get the logs and are looking at it. i will update as soon as we find out.

#8

admin
2,930 Posts
May 13, 2020, 3:26 am

Can you please try the below installation and please let us know if this fixes it.

https://drive.google.com/open?id=1d5-Q6XmKM3sT5d0aJZLw3Pd7GoP3pLC6

 

Last edited on May 13, 2020, 3:27 am by admin · #9

raf.schandevyl@telenet.be
6 Posts
May 13, 2020, 6:43 am

This install solved the problem. i can create bond's and set jumbo-frames for backend & iSCSI .

Thanks

Raf

#10
Pages: 1 2