VMware Cloud Community
sandsfootgroup
Enthusiast
Enthusiast
Jump to solution

Syslog Missing Server Hostname After vCenter 6.7 Update 1

Hello all

We have encountered an odd problem with syslog after upgrading vCenter 6.5u1g to vCenter 6.7u1.

Before the upgrade, vCSA was configured to forward syslog to a remote central syslog collector (UDP on port 514).  After we performed the upgrade from 6.5u1 to 6.7u1 the administrator of the remote central syslog system highlighted that vCSA was now sending mal-formed syslog messages.

We are still seeing some correctly formed syslog messages, but it appears that vCSA is also sending mal-formed syslog messages with a reverse format date instead of the vCSA hostname.  We performed some tests and monitored using tcpdump the messages that vCSA was sending.

This is the TCPDUMP taken when checking the syslog configuration in vCSA VMAI console (you can click to send a test message in the GUI so it is easy track).  We clicked it three times:

root@vcsa01 [ ~ ]# tcpdump dst x.x.x.x -ttttnvS | grep -i 'vcenter'

tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 262144 bytes

        Msg: 1 2019-01-18T12:07:38.358450+00:00 vcsa01 - - -  This is a diagnostic syslog test message from vCenter Server.\0x0a

        Msg: 1 2019-01-18T12:07:42.712657+00:00 vcsa01 - - -  This is a diagnostic syslog test message from vCenter Server.\0x0a

        Msg: 1 2019-01-18T12:07:43.494963+00:00 vcsa01 - - -  This is a diagnostic syslog test message from vCenter Server.\0x0a

380 packets captured

383 packets received by filter

0 packets dropped by kernel

The syslog message we are seeing that is mal-formed has the 'logged out' string inside it.  Again here is the TCPDUMP taken:

root@vcsa01[ ~ ]# tcpdump dst x.x.x.x -ttttnvS | grep -i 'logged out'

tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 262144 bytes

        Msg: 1 2019-01-18T12:09:30.271720+00:00 vcsa01 vpxd 5552 - -  Event [7561] [1-1] [2019-01-18T12:09:30.270543Z] [vim.event.UserLogoutSessionEvent] [info] [VSPHERE.LOCAL\vpxd-extension-d4ad4a50-d3eb-11e5-b37a-000c29a0ab87] [] [7561] [User VSPHERE.LOCAL\vpxd-extension-d4ad4a50-d3eb-11e5-b37a-000c29a0ab87@127.0.0.1 logged out (login time: Friday, 18 January, 2019 12:09:30 PM, number of API invocations: 2, user agent: VMware vim-java 1.0)]\0x0a

        Msg: 1 2019-01-18T12:09:43.774170+00:00 2019-01-18 12 - - - 09:43,775 - UserId : e671ee02-d6de-4431-bb8e-735a63623545, UserName : maintenanceAdmin, AuthSource : LOCAL, Session : e671ee02-d6de-4431-bb8e-735a63623545::cf71e003-c944-4f54-bee7-e23d43c51f7f, UserAction : LOGOUT,  - User logged out successfully\0x0a

        Msg: 1 2019-01-18T12:09:49.925903+00:00 2019-01-18 12 - - - 09:49,927 - UserId : e671ee02-d6de-4431-bb8e-735a63623545, UserName : maintenanceAdmin, AuthSource : LOCAL, Session : e671ee02-d6de-4431-bb8e-735a63623545::40a221f5-ab4c-46c1-a04b-00df80c8dd38, UserAction : LOGOUT,  - User logged out successfully\0x0a

        Msg: 1 2019-01-18T12:09:49.975762+00:00 2019-01-18 12 - - - 09:49,977 - UserId : e671ee02-d6de-4431-bb8e-735a63623545, UserName : maintenanceAdmin, AuthSource : LOCAL, Session : e671ee02-d6de-4431-bb8e-735a63623545::c42feb72-551b-49db-bcf3-640f54236498, UserAction : LOGOUT,  - User logged out successfully\0x0a

373 packets captured

373 packets received by filter

0 packets dropped by kernel

In this capture, you can see the first message is correctly formed.  Field one has the date/time stamp, field two has the hostname.

The next three messages are mal-formed.  Field one has the date/time stamp, but field two has date reversed, instead of the hostname.

This is causing us great headaches as we have several vCSA servers that will be upgraded and sending to a remote syslog server.  The issue being that the syslog server will not know the source of these messages.

As an aside to this, we also see vROPs 6.6.1 sending similar and then mal-formed syslog messages when sending to a remote syslog server.

Has anyone see this behavior before and any ideas of how to resolve?

Regards

Tony

0 Kudos
1 Solution

Accepted Solutions
sandsfootgroup
Enthusiast
Enthusiast
Jump to solution

Hello All

Well we managed to resolve this however, not sure why....

We saw similar entries - fields had reverse date rather than hostname generated by a vROPs appliance.  So we powered off the vROPs server and these vCenter mal-formed syslog messages stopped.

We also noticed that in our DEV environment we have the same version of vROPS but were not getting these vCenter mal-formed syslog messages.  We set about working out what could be different between these two environments discovered that the environment we were seeing mal-formed messages was using an upgraded to 6.6.1 vROPs appliance, but in DEV we had deployed a 6.6.1 vROPs ova. 

NB: vROPs had been upgraded to 6.6.1 a while before vCenter upgrage.

So we moved the original upgraded vROPs appliance sideways and deployed a new 6.6.1 from ova and bingo!  no more mal-formed syslog messages.

Odd but this worked for us.

Hope this helps

View solution in original post

0 Kudos
1 Reply
sandsfootgroup
Enthusiast
Enthusiast
Jump to solution

Hello All

Well we managed to resolve this however, not sure why....

We saw similar entries - fields had reverse date rather than hostname generated by a vROPs appliance.  So we powered off the vROPs server and these vCenter mal-formed syslog messages stopped.

We also noticed that in our DEV environment we have the same version of vROPS but were not getting these vCenter mal-formed syslog messages.  We set about working out what could be different between these two environments discovered that the environment we were seeing mal-formed messages was using an upgraded to 6.6.1 vROPs appliance, but in DEV we had deployed a 6.6.1 vROPs ova. 

NB: vROPs had been upgraded to 6.6.1 a while before vCenter upgrage.

So we moved the original upgraded vROPs appliance sideways and deployed a new 6.6.1 from ova and bingo!  no more mal-formed syslog messages.

Odd but this worked for us.

Hope this helps

0 Kudos