cruel
Programmer
- Aug 6, 2001
- 131
No time no see. I had couple of occasions with the following expreience:
I am running some jobs that involve, say, two parts: one is related to I/O, the other is calculation. It also involves looping over different units. The job skipped some units in the middle of the process. The second time it happened, I noticed that, while the server will not accept any new process, it was never shutdown. So, my job was never killed. I would like to confirm my speculation:
if the swaping space is stretching to the limit, it causes all I/O activities suspended. But the central processing never stopped running. So, in processing my job, it temporarily ignores I/O requests, causing skipping units. When working space is relieved, the I/O activities were resumed. Does this make any sense?
I am running some jobs that involve, say, two parts: one is related to I/O, the other is calculation. It also involves looping over different units. The job skipped some units in the middle of the process. The second time it happened, I noticed that, while the server will not accept any new process, it was never shutdown. So, my job was never killed. I would like to confirm my speculation:
if the swaping space is stretching to the limit, it causes all I/O activities suspended. But the central processing never stopped running. So, in processing my job, it temporarily ignores I/O requests, causing skipping units. When working space is relieved, the I/O activities were resumed. Does this make any sense?