#hoopla

My aim was to play a bit with Ceph to get familiar with the object store. Setting up Ceph is not very easy if something is different from the example in the docs. I don't want to use a whole block device for OSD for example.

I used the docker image https://hub.docker.com/r/ceph/demo/ for that. I installed and started it:

docker pull ceph/demo
docker run -d --net=host -v /etc/ceph:/etc/ceph -e MON_IP=192.168.0.20 -e CEPH_PUBLIC_NETWORK=192.168.0.0/24 ceph/demo

MON_IP has to be the own host's IP and CEPH_PUBLIC_NETWORK the corresponding own network. docker ps didn't show a running container afterwards, so I deleted the container again with docker rm CONTAINER-ID and restarted it without the -d. It showed that my harddisk space was not sufficient, so I added mon data avail crit = 1 to /etc/ceph.conf. docker run CONTAINER-ID worked now; docker ps shows the running container.

I tried to connect to the object store with the Rados Java binding. The import in the examples of the Ceph docs was not quite correct. The corrected version is:

import com.ceph.rados.Rados;
import com.ceph.rados.exceptions.RadosException;
import java.io.File;

public class CephClientTest {

public static void main(String args[]) {

try {
Rados cluster = new Rados("admin");
System.out.println("Created cluster handle.");

File f = new File("/etc/ceph/ceph.conf");
cluster.confReadFile(f);
System.out.println("Read the configuration file.");

cluster.connect();
System.out.println("Connected to the cluster.");

} catch (RadosException e) {
System.out.println(e.getMessage() + ": " + e.getReturnValue());
}
}
}


The first output was

Ceph cluster; ENOENT: No such file or directory (-2)


This was related to the permissions of /etc/ceph/ceph.client.admin.keyring. After adding read permissions to it for other users the connection could be established successfully.

I tried to configure an NTP server within a closed network without internet connection and make many clients synchronize with it (server and clients all on Ubuntu 16.04). Recipe:

• Install ntp on all nodes (ntpdate is not needed and should either be deinstalled or at least not be started)

• On the server: remove external NTP servers and make NTP use the local time. My /etc/ntp.conf only contains:

  server 127.127.1.0
  fudge 127.127.1.0 stratum 9
  

• service ntp restart; ntpq -p should give something like

       remote           refid      st t when poll reach   delay   offset  jitter
  ==============================================================================
  *LOCAL(0)        .LOCL.           9 l    4   64  377    0.000    0.000   0.000
  

• On the clients: My /etc/ntp.conf only contains:

  server NTP-SERVER iburst prefer
  

• service ntp restart; ntpq -p should give something like

       remote           refid      st t when poll reach   delay   offset  jitter
  ==============================================================================
   NTP-SERVER          .INIT.          16 u   31   64    0    0.000    0.000   0.000
  

I changed the hosting of my B2Evo blog to SSL. The only necessary configuration change was the $baseurl in conf/_basic_config.php. But there remained an issue with mixed content. Images of posts were still rendered by http, because the image links are hard-coded in the database. The following query helped:

UPDATE evo_items__item 
set post_content = REPLACE(post_content, 'http://www.fabianfier.de', 
'https://www.fabianfier.de') 
where post_content LIKE '%http://www.fabianfier.de%';

#tar cfvz xyz.tar.gz xyz 
tar (child): xyz.tar.gz: Cannot open: Disk quota exceeded
tar (child): Error is not recoverable: exiting now

HAHA: This machine pretends to use roughly twice as much inodes as the datatype long AND it says it has 500k free inodes:

# df -i
Filesystem                   Inodes                IUsed     IFree IUse% Mounted on
/dev/vzfs      18446744069909905553 18446744069415334917 494570636  100% /

#rm -rf xyz

Minutes(!) later:

#df -i
Filesystem     Inodes IUsed  IFree IUse% Mounted on
/dev/vzfs      750000 45285 704715    7% /

Within a controller I tried to redirect the user by return $this->redirect($this->generateUrl('...'));, but I couldn't find the corresponding route name. I did not define any baseroutepattern or baseroutename in the Admin file. In order to find out the route name, locate the service entry for the corresponding admin in admin.yml and copy the service name. On the terminal, type bin/console sonata:admin:explain and paste the service name to the end of the command. This command shows all routes associated with the service which can be used in generateUrl().