All my file servers at work (and at home) are Linux based... some are even using the same RH version you are using (7.3). (I'm too lazy to reboot them in-order to upgrade them to 9.0...) Moving GB size files is something I do on a daily basis and I've yet to see what you describe. (And my main file-server's uptime is passing the one year mark).
I'm not saying that you're wrong, I am saying that before you start a 'Linux sucks' advocacy thread, one would suggest that you start by reading more about the IDE controller that you are using (Intel chipset?) and the kernel support it has in 2.4.18. Next, do 'man hdparm' and check the current IDE support flags. (UDMA133 may require 'hdparm -c3 -u1 -X69 /dev/xxx' to reach optimal performance.) In general, people tend to ignore threads that are headed into troll-land. I assume that it was never your intention. -- Take care, Gilboa Davara XML - Systems Israel. mailto:[EMAIL PROTECTED] 972 - 054 968 909 On Fri, 2003-06-13 at 14:36, Eli Billauer wrote: > Hello again, > > Either most of you don't believe me, or you don't think that a Linux box > getting stuck is so serious. > > I have gotten some responses, some in private, with the same > experiences. Someone mentioned going to a coffee break when a large file > is being untarred (I do the same). And I ask you all: Isn't this Linux, > the operating system which is supposed to be the working horse? The OS > you trust when things get really tough? > > I did some testing of my own. If you want to skip the details, this is > the headline: Windows 2000 performed very well. Linux 2.4.18 got stuck. > > Description of the test: While my original complaint was regarding a 1.7 > GHz Celeron, all the following tests were performed on one single > computer, being a P4 2.4 GHz with a WD 80 GB hard disk, 845PE-based > Intel motherboard. > > It's not easy to measure the writing rate, since the throughput depends > on the state of the cache when the operation starts. But see remarks on > the Windows 2000 tests. > > In all cases, where there was a reasonable system response in general, > it was also fairly possibly to access the disk in parallel with the > large file write. Blaming the slowdown on bad disk access doesn't look > right to me. > > The test was performed by running "dd if=/dev/zero of=some_junkfile > bs=1M count=1000". The system response was then measures in a subjective > manner by performing GUI operations (opening and closeing windows, or > for that matter CTRL-SHIFT-Fx's, which didn't always react. Stuck means > stuck). > > Here is the summary of tests: > > Environment FS Result > ==================================================================== > RH 7.3 Linux 2.4.18-3 ext3 Totally stuck. CTRL-SHIFT-Fx > switches taking several seconds. > Difficult to stop process, because > ps runs very slowly > > RH 7.3 Linux 2.4.18-3 ext2 Same as ext3. The absence of > journal was double-checked (mount > didn't work as -t ext3, and > tune2fs reported absence of > has_journal flag) > > RH 7.3 Linux 2.4.18-3 FAT32 System slowdown, but still fairly > responsive > > DemoLinux 3.0(Deb.) 2.2.18 ext2 System slowdown, gets occasionally > stuck for short periods > > DemoLinux 3.0(Deb.) 2.2.18 FAT32 System slowdown, but nice response > > Cygwin under Windows 2000 FAT32 Windows responds slightly slower, > but even the Explorer is usable. > Writes at around 20 MB/sec, which was > 5-6 times *faster* than the 2.4.18 > results. > > Cygwin under Windows 2000 NTFS Grossly like FAT32 > ==================================================================== > > I'll leave you all to make your own conclusions. > > I'll be glad to get ideas of how to get Linux out of the mud, but even > if I succeed to get Linux as good as Windows 2000 (in this aspect), keep > in mind that almost anyone will judge Linux according to the performance > of what he or she gets out of the box of a mainstream distro. > > With sincere sadness, > Eli > > > > > > ================================================================= > 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] > ================================================================= 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]
