Servers Cold Start and Ram Module Error

Cold Start on Servers

Physical servers are likely to be upgraded at least once in their lifetime. For this reason, cold start is equally likely to be performed on servers. Compared to personal computers, upgrading is not something that can be done frequently because it is more difficult and risky. Especially if you have virtual servers and connected applications running on it, it will be very difficult.

Recently we decided to upgrade one of my VmWare Esxi servers to increase Memory. I stopped all running services and sent the applications to the cluster, but RAM boosting is not an easy process. Therefore, I wanted to share with you the problems I encountered.

First of all, I should say that servers work differently from personal computers. It can host 2 – 4 – 8 or 16 physical Processors. Each processor works with its own Ram module. The server I am using is HP DL380 G6 model. Although this model has a dual processor socket, it has a single processor on it.

The Ram Slots and Processor Sockets, which we show with numbers 1 and 2 on the open version of the server board below, work together. So ram module number 1 works with processor number 1.

The memories installed in the ram module are replicated within themselves. In other words, all of them must be the same model and speed. The model numbers written on the ram label must be exactly the same. The warning you will receive if even one of them is different is recorded in our video below. Approximately 6 seconds after the Cold Start process, a single and long audible signal indicates an alignment error in the Ram module or a defective Ram.

You have watched the video taken from 4 different angles of the test process we did with 9 pieces of memory in a single processor and a single ram module. The only problem with HP servers is that you cannot see which slots are incompatible.

As you can see in the first seconds of the video, all module LEDs on the front panel are lit. Your task is to check each ram one by one by eye and identify the model that does not fit in the slot, is defective or incompatible.

Cold Start

While we opened the inside of the server, we also recorded the “cold start”, that is, the “first start” event. In order to make Cold Start on servers, it is to connect the power cables and press the power key a few seconds after completely disconnecting the power cables.

If the server is running normally, you can also do this through ILO Management. The importance of cold start is that the fans run at full power for 3-4 seconds, so that the fans are tested for forcing and dusting inside the server is prevented.

There is also a Cold Restart option on the ILO manager. The function of this option is equivalent to physically pulling the cables. This reason should only be used in case of freezing or crashes in the operating system. Otherwise the platform may crash.