Re: Fwd: Where'd the cpu cycles go ?
3.5 orders of magnitude suggests he went from ram access (perhaps cache) to disk access. my guess is your configuration of 8.0 takes more memory = he's thrashing. what does free tells him in each case ? Hetz Ben Hamo wrote: Be careful with RH 8. 0 if you plan to use it on some heavy workload with some scripts... Thanks, Hetz -- -- regards +--- + Guy Baruch , Plasma Laboratory, Weizmann Institue. + mailto:guybar;plasma-gate.weizmann.ac.il + phone: 972-8-934-2211 +--- They hang the man and flog the woman That steal the goose from off the common, But let the greater villain loose That steals the common from the goose. -- English folk poem, circa 1764 http://bostonreview.mit.edu/BR27.3/bollier.html = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Fwd: Where'd the cpu cycles go ?
Well, the answer is a big different... If you run this script with en_US encoding and NOT utf8 - you would see the missing speed. Apparently there's a bug with grep when it comes to utf8 handling (redhat and Alan confirmed that).. So, if you're running lots of scripts on RH 8.0, maybe you should add a simple en_US encoding to the beginning of your script... Thanks, Hetz On Tuesday 05 November 2002 08:46, Gilad Ben-Yossef wrote: On Mon, 2002-11-04 at 23:10, Hetz Ben Hamo wrote: Be careful with RH 8. 0 if you plan to use it on some heavy workload with some scripts... Warning: This email is written before the first coffee of the morning. Please forgive me in advance... :-) I'm probably stupid (see above), but he didn't mention running it on the same *computer*. I mean, I would expect much more for this to be a result of some hardware related problem (HD operating with DMA off for example or some such) rather then some the wide VM problem it seems to be hinting at. In short, I think something is wrong in his configuratrion (and this may or may not be related to RH8.0) rather then to a bug somwewhere. Gilad = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Fwd: Where'd the cpu cycles go ?
On Tue, 5 Nov 2002, Hetz Ben Hamo wrote: Well, the answer is a big different... If you run this script with en_US encoding and NOT utf8 - you would see the missing speed. Apparently there's a bug with grep when it comes to utf8 handling (redhat and Alan confirmed that).. So, if you'rerunning lots of scripts on RH 8.0, maybe you should add a simple en_US encoding to the beginning of your script... Actually: 'en_US' is not guaranteed to exist. And is not guaranteed not to be UTF-8. 'C' will always be there. it is pure ascii. (And it is 4 letters shorter, which is why I have not suggested 'POSIX' ;-) ) -- Tzafrir Cohen mailto:tzafrir;technion.ac.il http://www.technion.ac.il/~tzafrir = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Fwd: Where'd the cpu cycles go ?
Be careful with RH 8. 0 if you plan to use it on some heavy workload with some scripts... Thanks, Hetz -- Forwarded Message -- Subject: Where'd the cpu cycles go ? Date: Monday 04 November 2002 13:30 From: Greg Hosler [EMAIL PROTECTED] Hi, I just came across a real performance bog in Red Hat 8.0 I noticed one of my sysadmin scripts going painfully slow, and running the cpu at 100%. upon some investigation, I noticed that a certain grep statement was choking the cpu. I've generated a shell script which duplicates that particular performance bog-down, but I can't pinpoint where the cpu's being hogged. The script generates a file (a.out) a little over a meg big, some 9 lines, and then times a grep thru the file. running this same script on Red Hat 7.3, I see times of (on average) .07 seconds to run the grep. On Red Hat 8.0, the same grep takes over 6 and a half minutes (no joke!). (i.e. a 4000 times difference, from under a tenth of a second, to almost 400 seconds) My first thought was what happened to grep, so I mounted my RH7.3 / partition on /mnt/root, and ran /mnt/root/bin/grep, and it also took the long path, so it's not grep. I ran a quick query in bugzilla (querying description and/or Summary for performance, and didn't see anything related. Can anyone reproduce (or void) my results ? -Greg +-+ You can release software that's good, software that's inexpensive, or software that's available on time. You can usually release software that has 2 of these 3 attributes -- but not all 3. | Greg Hosler [EMAIL PROTECTED]| +-+ --- grep_test Description: Binary data
Re: Fwd: Where'd the cpu cycles go ?
On Mon, 2002-11-04 at 23:10, Hetz Ben Hamo wrote: Be careful with RH 8. 0 if you plan to use it on some heavy workload with some scripts... Warning: This email is written before the first coffee of the morning. Please forgive me in advance... :-) I'm probably stupid (see above), but he didn't mention running it on the same *computer*. I mean, I would expect much more for this to be a result of some hardware related problem (HD operating with DMA off for example or some such) rather then some the wide VM problem it seems to be hinting at. In short, I think something is wrong in his configuratrion (and this may or may not be related to RH8.0) rather then to a bug somwewhere. Gilad -- Gilad Ben-Yossef [EMAIL PROTECTED] http://benyossef.com Geeks rock bands cool name #8192: RAID against the machine = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]