96

The Three A.M. Connection

새벽 세 시의 접속

  • 3,590 characters
  • ~11 min

February 22. I worked the night for the month-end settlement batch.

The computer room keeps people back at month end. The settlement batch runs at night and if it fails partway it has to be started again. Two of us stayed that night. Me and one batch operator. He went to get his head down after twelve and I stayed. There was not much for me to do, but it was my first time so they said to watch.

At two I watched the backup start. A progress figure came up on the screen and the number climbed. At half past two the cleanup started. At three the indexing started.

Seeing the hours I had read in the log actually happen was nothing at all. A line comes up on the screen and then the next line does.

An office at three in the morning sounds different from an office by day. With no people in it the machines are loud. The sound past the server-room door is background by day and at night it is the only thing. I listened to it a long while and noticed it has no rise or fall. One tone held long enough stops sounding like a sound and starts sounding like the state of the room.

A little after three I opened the session list. No particular reason. Working nights, you open screens you never open otherwise.

The session list shows the seats currently attached to the company network. At three in the morning there should be none. There were two lines.

One line was my seat. My staff number sat at the right of the screen.

The other line was the archive terminal.

I looked at that line and for a moment thought nothing. More exactly, the thought took a while to come round. The archive was emptied on February 8. Three terminals stayed as assets and I dropped their power myself. The key went back to facilities.

Waking the operator seemed excessive, so I decided to ask in the morning and opened a few more screens instead.

The session information had a connection start time. Two fifty-nine. One minute before the indexing job starts.

What that seat is doing does not appear in the session list. That is a different screen and needs a privilege. I did not have it.

At three forty-one I refreshed the list and the line was gone. Only my seat was left.

I opened the connection log. It was there. Start two fifty-nine, end three forty-one. For forty-two minutes, the record said, the archive terminal was attached to the company network.

In the morning I asked the operator. He answered over his coffee. If the power is down it does not show. So what is this, I said, and showed him the log. He looked briefly and asked whether it had gone into sleep. In sleep the screen goes off and the line stays live, he said.

I said again that I dropped the power. That I put the switch down.

Then that is odd, he said. He said it was odd and went off to his own work. Odd did not go any further than that. Odd things in a computer room usually end that way. If it does not reproduce, no one looks for the cause.

I thought he was right. Chasing what does not reproduce means the work does not get done. Except this one was not failing to reproduce; it was simply not tried. To check it, you go down to the basement.

In the afternoon I borrowed the key from facilities.

The second basement after two weeks. The air was cold when the door opened. The heating is off down there, so of course. With the lights on, six rows of bare shelving and three desks stood as they were.

The three terminals were as they were too. I checked them one by one. The power switches were down. All three.

I put the back of my hand on a screen. The glass was cold. I put a hand to the vent at the back. Cold there too.

Touching them, they were cold. A machine that has run all night is lukewarm in the morning. These were at the temperature of machines that had been off for two weeks.

I checked the lines as well. Opening the junction box on the wall, all three wires were in. Being plugged in and being switched on are different things. With the wire in and no power, no signal goes anywhere. Anyone who used a terminal at a branch knows that much.

Before leaving the room I looked at the square mark on the desktop. The new terminal had not been moved, so the mark was still under the machine. Only its border showed.

Returning the key, I asked the facilities man whether anyone had been into that room lately. He turned the register over and said no one since February 8. Mine was the first key taken out, today.

Back upstairs I thought for a while about how to write it up.

Written plainly, the log would say that a machine with its power down was attached to the network for forty-two minutes. Written that way, someone will read it later and ask. Asked, I have nothing to answer with. Two things are confirmed: that the log says so, and that the machines are off and cold.

I wrote both. I did not write a judgement. When there is no ground for a judgement it is better to write only what was seen — that I learned at the branch. If the cash does not balance you write that it does not balance; you do not invent why.

Having written it I raised the question of privileges with the man beside me. What privilege lets you see what a session is doing? Operations, he said, and that has to be applied for with a team head's approval. What would you put as the reason, he asked.

I could not answer. No sentence for the reason field came to me. Write that a machine that is off attaches every night and you have to explain that first, and explaining it comes down to one line of log and three cold machines.

Closing the screen, I opened the connection log once more.

The previous night's record was there. The night before that too. I went back through the dates.

Since February 8, every night, there had been a session for the archive terminal. The start was usually two fifty-something and the end in the three forties.

Fifteen days, every one. I saw it today for the first time.

I went further back. Before February 8 there was no such line. Before that there were only daytime connections, and those are the hours I was working in the archive.

So the session moved to the night from the day I emptied that room. While a person was there it attached by day, and once no one was there it attaches at night.

Whether to put that sentence in the log I thought about for a long time. Written, it is a judgement; unwritten, it is something I did not see. I did not write it. Instead I printed the connection log and put it in a drawer. On paper it can be looked at again later.

Comments

  • No comments yet. Be the first.

Sign in to join the conversation.