Skip to content

Conversation

@donald
Copy link
Collaborator

All done, PR just for documentation

The individual netlog services send messages from system logfiles (e.g.
/var/log/messages), which they consider interesting, to the clusterd of
a specific node. The receiving clusterd uses syslog() to log these
messages, so they end up in /var/log/message of that node.

Let the clusterd of 'macheteinfach' listen to messages from the netlog
services instead of that on 'lol'. We use 'macheteinfach', because the
only current consumer of these messages is /project/azubi, which happens
to run on 'macheteinfach'. This way, this project no longer needs to use
ssh to get to the right logfile.
Send 'interesting' messages to the clusterd on macheteinfach instead to
that one on 'lol'. See previous commit.
Each clusterd sends regular status messages (version and load
information) to a single clusterd which currently just collects them for
operator review.

    buczek@macheteinfach:~/git/mxtools (phaseout-lol)$ telnet lol 234
    Trying 141.14.29.220...
    Connected to lol.
    Escape character is '^]'.
    clusterd 20220118-124724 stupid console
    For historical messages, grep "clusterd" from /var/log/messages on lol (or "tail -f /var/log/messages |grep cluster")
    v
            abelson : ['DOWN',0,28,'0.01','V1.108 - 20170215-151907',0,'434420.8','4.9.38.mx64.164']
                ace : ['UP','95647416.84',1,0,'20220118-124724',0,'921594.24','5.10.70.mx64.403']
             acedia : ['UP',0,1,'0.72','20220118-124724','0.00104166666666667','30613821.35','5.10.24.mx64.375']
                afk : ['UP','95647391.2',1,'0.1','20220118-124724',0,'5941821.44','4.14.87.mx64.236']
       agentwhiskey : ['UP',0,1,'0.07','20220118-124724',0,460800,'5.10.82.mx64.414']
    [...]
    Connection closed by foreign host.

The utility of this can be debated and maybe we should remove that
completely.

Anyway, for now send status messages to "afk" instead of "lol", because "lol" is
going to be taken out of commission.
Some of the communications of clusterd is based on ip broadcasts. If the
nodes are distributed over multiple ipv4 networks, different nodes need
to do the broadcasts, because without special router setups, you can't
broadcasts to remote subnets.

For this reason, one node is a designated broadcaster for each subnet.
This node is called "area router" in the code. Data to be sent to all
nodes is sent via unicast to each area router which, in turn, brodcast
it to the networks they are resposible for.

Move the area router role for network 141.14.16/0 with broadcast address
141.14.31.255 from lol to wtf.
Sign in to join this conversation on GitHub.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant