Speed problem in a new install of 9.3
Not sure where to look to try and figure out what the problem is. I was running SuSE9.3 on a P3 1GHz with 1GB RAM and an ATI 7000 video card... standard IDE hard drive. Onboard sound... ummm.. Realtec NIC. Nothing special.. things ran reasonably well on it - SuSE9.3 basic 32 bit install, with VMWare, Oracle10g and a Java based app or two. It finally died and I got a new machine. The new one is an AMD64 3000+ with 1GB RAM, and an ATI 9600 video card. Now I have a SATA drive instead of IDE... onboard sound/NIC etc. Same 9.3 32-bit install (YOU updated of course) with the same exact VMWare images, Oracle10g and the Java apps. I've had to go with the 32bit install on the 64bit machine because of the Oracle version I have to work with and a few other commercial/custom apps that are 32 bit only (the long list of dependancies the apps have means I have to uninstall most of the 64 bit install and install 32 bit versions... not worth the hassle). Anyway, I'm finding that the new machine - with in theory should be significantly faster than my old machine is in fact... slower than the old machine. I can't find out what's making it slower either. I've been monitoring my processes, and nothing out of the ordinary - that I can see anyway. But... my processor load sometimes pegs at 100% for no apparent reason - X for example taking 100% for minutes at a time.... but nothing is happening on screen... I'm not doing anything either (reading a text file... nothing processor/GUI intensive). There are times when I simply cannot do anything at all until the OS finishes whatever it is that's keeping it busy and returns control to me. Any suggestions where I can look or what I can monitor to try and find what it is that's making the system so slow? I do have one odd (repeating entry) in my /var/log/messages... ----------- Jun 24 06:36:42 smaug nmbd[6462]: find_domain_master_name_query_fail: Jun 24 06:36:42 smaug nmbd[6462]: Unable to find the Domain Master Browser name SMAUG<1b> for the workgroup SMAUG ------------ Jun 24 06:21:42 smaug nmbd[6462]: Unable to sync browse lists in this workgroup. Jun 24 06:36:42 smaug nmbd[6462]: [2005/06/24 06:36:42, 0] nmbd/nmbd_browsesync.c:find_domain_master_name_query_fail(353) ----------- This seems related back to a Samba share I have... I think, but I'm not sure since I don't know what is actually going wrong. Well, like I said, any suggesions would be more than welcome. C
On Fri, 24 Jun 2005 05:24 pm, Clayton wrote:
There are times when I simply cannot do anything at all until the OS finishes whatever it is that's keeping it busy and returns control to me.
I had problems like this with 9.2, never got to the bottom of it but worked around it by disabling a 2 gig swap partition. Back in 8.1 there was a disk tuning needed to tell it to hold less buffers and flush them less often. That system would become unresponsive for 2 minutes at a time. FWIW michaelj PS: My apologies to the SLES list, I accidentally cross posted. PPS: On the 9.2 workstation, it was an IDE drive on 8.1 it was a SCSI based server. I think DMA was turned on, why would it be turned off. -- Michael James michael.james@csiro.au System Administrator voice: 02 6246 5040 CSIRO Bioinformatics Facility fax: 02 6246 5166 No matter how much you pay for software, you always get less than you hoped. Unless you pay nothing, then you get more.
On Fri, 24 Jun 2005 18:24:11 +1000 Michael James <.> wrote:
On Fri, 24 Jun 2005 05:24 pm, Clayton wrote:
There are times when I simply cannot do anything at all until the OS finishes whatever it is that's keeping it busy and returns control to me.
Maybe SUSE's daily cronjobs? Doing "updatedb" takes a while on a system with larger disk-space... You could do e.g. 'ls -ltra /var/spool/cron/lastrun/' as root, to see if any of the cronjobs have just run. I mean after this "strange activity" have been finished once. Pelibali PS. I have SUSE 9.1, so it could be, that while using a newer version (~9.3) you have the above folder by default _somewhere else_.
On Friday 24 June 2005 09:24, Clayton wrote:
Anyway, I'm finding that the new machine - with in theory should be significantly faster than my old machine is in fact... slower than the old machine. I can't find out what's making it slower either. I've been monitoring my processes, and nothing out of the ordinary - that I can see anyway. But... my processor load sometimes pegs at 100% for no apparent reason - X for example taking 100% for minutes at a time.... but nothing is happening on screen... I'm not doing anything either (reading a text file... nothing processor/GUI intensive).
There are times when I simply cannot do anything at all until the OS finishes whatever it is that's keeping it busy and returns control to me.
Any suggestions where I can look or what I can monitor to try and find what it is that's making the system so slow?
Do you use konqueror? I see from time to time konqueror go completely mad, I think it has something to do with flash. When this happens, the CPU usage hits 100%, but it shows in different places. Sometimes in X, sometimes as "system" use. But invariably it goes away when I kill konqueror
Anders Johansson wrote:
On Friday 24 June 2005 09:24, Clayton wrote:
Anyway, I'm finding that the new machine - with in theory should be significantly faster than my old machine is in fact... slower than the old machine. I can't find out what's making it slower either. I've been monitoring my processes, and nothing out of the ordinary - that I can see anyway. But... my processor load sometimes pegs at 100% for no apparent reason - X for example taking 100% for minutes at a time.... but nothing is happening on screen... I'm not doing anything either (reading a text file... nothing processor/GUI intensive).
There are times when I simply cannot do anything at all until the OS finishes whatever it is that's keeping it busy and returns control to me.
Any suggestions where I can look or what I can monitor to try and find what it is that's making the system so slow?
Do you use konqueror? I see from time to time konqueror go completely mad, I think it has something to do with flash. When this happens, the CPU usage hits 100%, but it shows in different places. Sometimes in X, sometimes as "system" use. But invariably it goes away when I kill konqueror
There is also that find and slocate routine that starts up after your machine is idling. It is busy finding all the text and making a search index. control alt f1 logon root and killall slocate and killall find when you need to. Adam
On Tuesday 28 June 2005 22:07, Adam Vazquez Kb2Jpd wrote:
There is also that find and slocate routine that starts up after your machine is idling. It is busy finding all the text and making a search index. control alt f1 logon root and killall slocate and killall find when you need to.
If updatedb is seriously affecting your performance (especially on a modern 64 bit machine) then you have a DMA configuration problem. I would suggest you make sure you have it enabled on your hard drives
On 6/29/05, Anders Johansson
On Tuesday 28 June 2005 22:07, Adam Vazquez Kb2Jpd wrote:
There is also that find and slocate routine that starts up after your machine is idling. It is busy finding all the text and making a search index. control alt f1 logon root and killall slocate and killall find when you need to.
If updatedb is seriously affecting your performance (especially on a modern 64 bit machine) then you have a DMA configuration problem. I would suggest you make sure you have it enabled on your hard drives
I'm not using Konq much (Firefox as a browser). I sometimes have it open in filemanager mode... but not that often. I don't see slocate running, and updatedb also doesn't seem to be the culprit - if it is, it's running every 30 seconds... which it's not. DMA.... as far as I know all SATA drives use DMA by default. I also don't see anything when I look at the cronjobs. I get the CPU spikes even if Oracle is shut down and killed. I'll have to keep hunting I guess... frustrating though since I find that my new AMD64 3000+ work computer is slower than my old 2.2GHz Celeron at home (both running SuSE9.3). :-( C.
participants (5)
-
Adam Vazquez Kb2Jpd -
Anders Johansson -
Clayton -
Michael James -
pelibali