| Title: | Alpha Developer Support |
| Notice: | [email protected], 800-332-4786 |
| Moderator: | HYDRA::SYSTEM |
| Created: | Mon Jun 06 1994 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 3722 |
| Total number of notes: | 11359 |
Company Name : BGS Systems
Contact Name : Yefim Somin
Phone : (617) 663-4533
Fax :
Email : [email protected]
Date/Time in : 13-FEB-1997 15:21
Originally entered by : SWIERKOWSKI
SPE center : PAG
Category : Digital UNIX
OS Version : V3.2C
System H/W : AlphaStation 200 4/166 (256MB)
Brief Description of Problem:
-----------------------------
Has a couple of questions:
1) Wants to know about what processes are running on the system?
- Wants to know about nature of pid 0 and what it is doing/running?
- Why is "TIME" column non-zero? (and fairly large to boot!)
- What is meaning of "CMD" column set to: "[kernel idle]"?
2) Wants to know about global numbers of page in and page out operations (via
vmstat and/or table calls)?
- Why no disk activity showing up when paging activity is happening?
- What do disk I/O metrics really mean?
I spoke briefly with Yefim about his questions and tried to narrow down
what information/definitions he was after, but gave up after awhile. I've
paraphrased the basic issues/concerns above (I hope!). I went through the
man pages for ps, vmstat, and iostat, but I'm going to punt on this. It
really sounds like he needs to have an architecture/internals discussion
with someone, cheers...
Tony Swierkowski
Digital Equipment Corporation
Software Partner Engineering
Palo Alto, California
(415) 617-3601
"[email protected]"
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 3196.1 | More voicemail/email tag, no answers yet... | AMCUCS::SWIERKOWSKI | Quot homines tot sententiae | Mon Feb 24 1997 20:28 | 17 |
Greetings! Both myself and Jeff Smith have either spoken with and/or exchanged mail with Yefim about this and I understand the details a bit better of what his concerns are. I'll craft a note for the DIGITAL_UNIX conference tomorrow about this as all the research and reading I've done last week and today in COMETS, assorted notes conferences and the Internals SG I found on the net don't explain the metrics he is seeing. I'm going to leave this open awhile longer... Tony Swierkowski Digital Equipment Corporation Software Partner Engineering Palo Alto, California (415) 617-3601 "[email protected]" | |||||
| 3196.2 | [kernel idle] isn't, Call closed anyway... | AMCUCS::SWIERKOWSKI | Quot homines tot sententiae | Tue Mar 18 1997 19:22 | 26 |
Greetings! After much research on this, more email with Yefim and a detailed note thread (i.e. #8963.*) in the DIGITAL_UNIX conference, we both understand better what is going on. Specifically, the [kernel idle] process (i.e. PID 0) isn't really just a "thumb twiddler". Yefim did get back to me with a suggestion to get the [kernel idle] process renamed since it really is involved in assorted "system overhead" functions. He also asked about the per-thread activity displayed by 'ps mlp0' (especially "netisr" since that thread is the largest CPU consumer). Most of the threads displayed by 'ps mlp0' don't have names but numbers under the WCHAN heading and I don't know how to resolve this into meaningful routine names or whatever. Even the threads that have symbolic names (like "netisr") don't tell me much. I suspect that in order to take this much further, some time is going to have to be spent in the source listings. For now I'm going to close this call, cheers... Tony Swierkowski Digital Equipment Corporation Software Partner Engineering Palo Alto, California (415) 617-3601 "[email protected]" | |||||