Details
-
Bug
-
Status: Confirmed (View Workflow)
-
Critical
-
Resolution: Unresolved
-
23.10.3
-
None
-
None
-
2025-2
Description
cmapi writes on debian cluster 127.0.1.1 instead real address
Example
<ModuleIPAddr1-1-3>127.0.1.1</ModuleIPAddr1-1-3> |
<ModuleHostName1-1-3>dnsname</ModuleHostName1-1-3> |
|
<ClusterManager>127.0.1.1</ClusterManager> |
root cause is an entry in /etc/hostnames, which will be written on default on debian
Explanation:
https://www.debian.org/doc/manuals/debian-reference/ch05.en.html#_the_hostname_resolution
127.0.0.1 localhost |
127.0.1.1 host_name |
Test:
Create file gethostid.py
import sys, os |
import optparse |
import socket |
remoteServer = sys.argv[1] |
remoteServerIP = socket.gethostbyaddr(remoteServer)
|
print (remoteServerIP)
|
~$ python gethostid.py hostname
|
('hostname', [], ['127.0.1.1']) |
on non-debian systems real IP will be provided
Even it works , if the dns name will be locally used,
but columnstore.xml will be copied from node1 to other nodes,
so other nodes get a problem, if they use
<ModuleIPAddr1-1-3>127.0.1.1</ModuleIPAddr1-1-3> |
for PM1 as an IP.
So cmapi getting IP on debian systems in another way.
Config
/etc/hosts
can be a workarround, but due the fact, that is the default debian configuration,
it can be overwritten.
julien.fritsch since Richard lowered S1 to S3, I believe we can lower priority from blocker