The further my stay in Belfast progresses, the later my posts are written… I will try to summarise my week faster – for the last two weeks.
Tim and I continued working on the server changeover and finally exchanged the certificates of many machines. Especially for companies with a large internet presence, these certificates are very important because the website is closely linked to the product. When a certificate has expired or incorrectly configured, visitors to the website get the message “not secure”, which leads to them not using the product or at least having more doubts about it. Previously, we had mainly worked in the test environment. So the pressure was pretty high when we had to work with the production systems and of course we didn’t want to be responsible for the failure of the systems or even for a customer to jump off.
Luckily, almost everything went smoothly – almost… The servers in the test environment were quickly redeployed. But when we came to the website, the company’s flagship, probably the most important point of the conversion, it was getting tough for the first time.
Only three tasks had to be done:
- Run a script that fetches the new certificate.
- Change the server configuration to use the new certificate.
- Restarting the local Web server
The user does not have to deal with the changeover in most cases because the server always responds with a cached configuration. So if you change it, the new configuration will not be active until you restart the web server. This takes less than a second. Only at the second step something can go wrong, as 1 and 3 run completely automatically and even if something does go wrong, you should of course create a backup of the configuration as with any changeover. If the web server is in the 3rd Step and does not start again, you simply import the backup and then take your time to solve the problem. We had saved the website as the most important server for last, so that we could intercept all possible errors before. Even the 2nd Step, changing the configuration was almost automatic, because we simply have to copy the three rows to be changed from a working configuration and rewrite them slightly. So everything went through as before, but then the fright of the day: the Web server did not restart and displayed an error message: “Configuration faulty. Could not restart. ” The website was offline. 
“Import the Backup” – I, still calm.
“I have no backup” – Tim, no longer quite so quiet.
“Take a quick look at the config and correct the error. is probably just a typo. ” – I, already slightly panicked.
We went to the configuration and then came the shock: we did not see the error. In its error message, the Web server even indicates exactly the line in which it sees a problem, but in comparison to the already working configurations everything looked the same. We stared at the configuration for half a minute and looked for something that could be a problem – without success. Slowly, we both panicked. Every second in which we did not find the error, a customer could try to access the website and would not see anything. The Web server would not respond and its browser would display an error message.
We went back to the error message of the Web server. The problem should be caused by some strange characters in the configuration, which we could not see at all when looking at the configuration. Suddenly we realised what went wrong: When copying the configuration Tim had transferred two lines at once. However, characters were copied that did not belong to the configuration, but have to do with the window through which we access the other server with the working configuration.
But we could not see these letters – they were so far right in the line that they were not shown to us at all. That’s why they hadn’t noticed us. So Tim quickly deleted those characters. We quickly went over the configuration again so that nothing could go wrong anymore. We did not want to risk having an error message again when restarting and the page to remain offline for even longer. A maximum of 2 minutes had elapsed and probably no one noticed that something was wrong. So Tim gave the command to reboot and…. It worked!
The website was usable and showed the newly created certificate. Phew, at last the tension could fall off. The internship had been very relaxed until that time, so sometimes you could talk about boredom, but at that moment we were quite happy that not every day was so exciting.
At the end of the week Kristina came to visit me for the second and last time. We had a lot of nice things planned for the weekend and booked a different place to stay for the last time. That’s why we were in another area. This turned out to be a very good decision as we were right on the river Lagan and at Ormeau Park. When we looked out the window, we could see the river right behind a green meadow. For Kristina the highlight of the accommodation was the dog of the landlady. In the beginning he was very shy and barely moved away from his couch in the absence of the mistress, but after a lot of persuasion by Kristina, he sat on her lap on the second day.
On Thursday we went for dinner at Lavery’s to get some nice food and on Friday we tried to find a pub where you could enjoy traditional Irish music. Unfortunately, we were not particularly successful. Mostly the live music started very late and we had big plans the next day and we didn’t want to be out all night.
On Saturday we drove to Dublin together! Kristina wanted to look at more of the city than the airport and I wanted to go eat at the delicious restaurant from last time again. Unfortunately Kristina was no longer feeling so good. We still looked at a lot of the city and could do the most important things on my todo list (eat in the delicious restaurant and buy donuts – it may be that my todo list had fewer entries than Kristina’s).
On Sunday, Kristina was so ill that we couldn’t do anything and we had a relaxed day in our accommodation.
Only two more weeks are left in Belfast. So I have to worry about what else I want to see/do while I’m here!
