Re: [ilugd] New machine, minor irritants
I've played about with the disk scheduler and the buffering ratio (/sys/block/sda/queue/scheduler and /proc/sys/vm/dirty_background_ratio) and that seems to alleviate the problem a bit, but it's still there, albeit much less than before. On the other hand, I could be totally wrong about the cause of the problem, and this could be mere symptomatic treatment. So anyone have a clue as to why these freezes happen, or a better solution for fixing them? The disk is otherwise writing at about 60MB/s (which I presume is OK for a SATA). See if you have disk cache turned off. If so turning ON disk write cache should alleviate this problem but you better have your box on UPS. The other weird thing is the temperature sensors. CPU temperatures show up within limits (typically 55C plus/minus 5C), but two of the temperatures are way out of whack: AUX Temp: +127.0°C (high = +80.0°C, hyst = +75.0°C) ALARM sensor = thermistor Sys Temp:+74.0°C (high = +17.0°C, hyst = +43.0°C) ALARM sensor = diode When I first saw my AMD CPU temp at 70 C I was alarmed too but I think it is normal for CPUs to have temp this high, since CPU is the hottest thing in the box, may be that is what your Sys temp is. No idea about AUX. Regards, --Naresh Narang ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Raj Mathur writes: [snip] My AMD64 desktop was on its last legs, so I got myself 1/2 a new system -- Intel DG31PR motherboard, 4GB RAM, Core 2 Duo at 2533 MHz, CoolerMaster cabinet. Being cheap, I cannibalised my old hard disks and SMPS. System booted up first shot with the old Debian kernel, and except that somehow eth0 had got renamed to eth1 udev(7) is responsible. To fix the issue, edit '/etc/udev/rules.d/??-persistent-net.rules' to suit you. (and X didn't work since it was configured for the old Radeon 7250 card), it just recognised everything out of the box. Amit Kalra, my hardware supplier, who had been anticipating a week-long odyssey of reconfiguration for the new hardware didn't say anything but I could tell by the way his eyes grew to twice their normal size as the system was booting that he was pretty damn impressed. When 10 minutes of reconfiguration got the 'net and the graphics too fixed, his eyes moved into Quad size mode :) Go on Winduhs, do that and show us! Quad size mode ? Now one of the things I'm facing is a slowdown of the system when it's doing disk-write-intensive activities. I believe this is because of the huge amount of RAM -- Linux buffers disk writes, and when it does start flushing the buffers to disk everything else freezes. Firefox, e.g. freezes for up to 20 seconds when the writes are in progress. I've 2 GiB RAM, 2 SATA disks. And I've not expected this issue, unless I copy some big files and then execute 'sync' :) . Which kernel version are you running anyways ? May be your SATA controller is being used in some kind of compatibility mode. Debian recently released a new kernel for new hardware[1], may be shifting to that will fix the issue. I've played about with the disk scheduler and the buffering ratio (/sys/block/sda/queue/scheduler and /proc/sys/vm/dirty_background_ratio) and that seems to alleviate the problem a bit, but it's still there, albeit much less than before. On the other hand, I could be totally wrong about the cause of the problem, and this could be mere symptomatic treatment. So anyone have a clue as to why these freezes happen, or a better solution for fixing them? The disk is otherwise writing at about 60MB/s (which I presume is OK for a SATA). With 4 GiB of RAM, I'm wondering how much disk is it caching. I hope your swap is not dirtied by it, yet. The other weird thing is the temperature sensors. CPU temperatures show up within limits (typically 55C plus/minus 5C), but two of the temperatures are way out of whack: AUX Temp: +127.0°C (high = +80.0°C, hyst = +75.0°C) ALARM sensor = thermistor Sys Temp:+74.0°C (high = +17.0°C, hyst = +43.0°C) ALARM sensor = diode OK, the AUX is probably just some hardware or configuration glitch, since it's constant at 127C. However I'm a bit concerned about the Sys temperature -- should it be 70C+ ? Could it be a wrong reading, or do I need to do something to fix this? These are mines, on idle use :). CPU Temp:+68°C (low = -127°C, high = +127°C) Board Temp: +51°C (low = -127°C, high = +127°C) Remote Temp: +49°C (low = -127°C, high = +127°C) What /is/ the Sys temperature anyway? % fgrep System /etc/sensors.conf label temp2 System label fan2System To confirm which one is yours go through that file, and refer to the section corresponding to I2C chip your box is having. Any help, pointers appreciated. sensors output available on request. References: [1] - http://www.debian.org/News/2008/20080726 HTH - -- ·-- ·- ·--- ·- ···- ·- ·--·-· --· -- ·- ·· ·-·· ·-·-·- -·-· --- -- () ascii ribbon campaign - against HTML e-mail /\ www.asciiribbon.org - against proprietary attachments -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.9 (GNU/Linux) iEYEARECAAYFAkjE2xwACgkQHy+EEHYuXnSNNgCdGNdBOOeKxO4Gkd8wXtCKtrUn ZKMAoI8FogDqXJR5V6jX9UOvDCB+pKgP =vTGz -END PGP SIGNATURE- ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
Raj Mathur writes: System booted up first shot with the old Debian kernel, and except that somehow eth0 had got renamed to eth1 Ashish Shukla आशीष शुक्ल [EMAIL PROTECTED] writes: udev(7) is responsible. To fix the issue, edit '/etc/udev/rules.d/??-persistent-net.rules' to suit you. Yebbut udev persistence only came in on etch. Unless Raj is still running sarge? PJ ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
On Monday 08 Sep 2008, Naresh Narang wrote: I've played about with the disk scheduler and the buffering ratio (/sys/block/sda/queue/scheduler and /proc/sys/vm/dirty_background_ratio) and that seems to alleviate the problem a bit, but it's still there, albeit much less than before. On the other hand, I could be totally wrong about the cause of the problem, and this could be mere symptomatic treatment. So anyone have a clue as to why these freezes happen, or a better solution for fixing them? The disk is otherwise writing at about 60MB/s (which I presume is OK for a SATA). See if you have disk cache turned off. If so turning ON disk write cache should alleviate this problem but you better have your box on UPS. Hmm, that's an idea. However, according to sdparm, cache is turned on: Caching (SBC) mode page: WCE 1 RCD 0 One issue could be that earlier (with 1GB of RAM) the amount of writes that Linux was caching was enough to fit into the SATA cache, whereas now with 4GB it's more than what the cache will handle. Now I need to test out this hypothesis by first figuring out how much cache the SATA drive has... any clue? All dmesg says is that the cache is enabled, doesn't actually tell you the size. Sys Temp:+74.0°C (high = +17.0°C, hyst = +43.0°C) ALARM sensor = diode When I first saw my AMD CPU temp at 70 C I was alarmed too but I think it is normal for CPUs to have temp this high, since CPU is the hottest thing in the box, may be that is what your Sys temp is. No idea about AUX. Nah, CPUs, like I said, are at 55C +/-5C (depending on how many bzip2's I'm running to test the temperatures with :) . Sys is something else... Regards, -- Raju -- Raj Mathur[EMAIL PROTECTED] http://kandalaya.org/ GPG: 78D4 FC67 367F 40E2 0DD5 0FEF C968 D0EF CC68 D17F PsyTrance Chill: http://schizoid.in/ || It is the mind that moves ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
On Monday 08 Sep 2008, Ashish Shukla आशीष शुक्ल wrote: Raj Mathur writes: [snip] Now one of the things I'm facing is a slowdown of the system when it's doing disk-write-intensive activities. I believe this is because of the huge amount of RAM -- Linux buffers disk writes, and when it does start flushing the buffers to disk everything else freezes. Firefox, e.g. freezes for up to 20 seconds when the writes are in progress. I've 2 GiB RAM, 2 SATA disks. And I've not expected this issue, unless I copy some big files and then execute 'sync' :) . Which kernel version are you running anyways ? May be your SATA controller is being used in some kind of compatibility mode. Debian recently released a new kernel for new hardware[1], may be shifting to that will fix the issue. Running 2.6.26 (which I believe is the latest). And yes, the problem is exactly what you face when copying big files and then running sync, except in this case it's writing large data and then freezing when the internal, automatic cache flush happens. [snip] % fgrep System /etc/sensors.conf label temp2 System label fan2System To confirm which one is yours go through that file, and refer to the section corresponding to I2C chip your box is having. Hmm, I think I'll take this to the lm-sensors mailing list -- could be a bug in sensors. Have also asked Amit to check the corresponding temperatures on Winduhs machines using the same motherboard, let's see if that gives us any information. Regards, -- Raju -- Raj Mathur[EMAIL PROTECTED] http://kandalaya.org/ GPG: 78D4 FC67 367F 40E2 0DD5 0FEF C968 D0EF CC68 D17F PsyTrance Chill: http://schizoid.in/ || It is the mind that moves ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
On Monday 08 Sep 2008, PJ wrote: Raj Mathur [EMAIL PROTECTED] writes: Now one of the things I'm facing is a slowdown of the system when it's doing disk-write-intensive activities. I believe this is because of the huge amount of RAM -- Linux buffers disk writes, and when it does start flushing the buffers to disk everything else freezes. Firefox, e.g. freezes for up to 20 seconds when the writes are in progress. Did vmstat suggest lots of page outs at the same time? Didn't look. Will have to reset the tuning I've done, run recollindex and check. Did you try lowering swappiness ? Would that help? I doubt if swap is the issue here (all programs fit fine into memory, and there's 2.5G+ of memory available after discounting buffers). IAC, will test again after untuning :) Regards, -- Raju PJ ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/ -- Raj Mathur[EMAIL PROTECTED] http://kandalaya.org/ GPG: 78D4 FC67 367F 40E2 0DD5 0FEF C968 D0EF CC68 D17F PsyTrance Chill: http://schizoid.in/ || It is the mind that moves ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
See if you have disk cache turned off. If so turning ON disk write cache should alleviate this problem but you better have your box on UPS. Hmm, that's an idea. However, according to sdparm, cache is turned on: Caching (SBC) mode page: WCE 1 RCD 0 One issue could be that earlier (with 1GB of RAM) the amount of writes that Linux was caching was enough to fit into the SATA cache, whereas now with 4GB it's more than what the cache will handle. Now I need to test out this hypothesis by first figuring out how much cache the SATA drive has... any clue? All dmesg says is that the cache is enabled, doesn't actually tell you the size. Cache is usually small and kernel is supposed to handle that part, I don't think it can be increased. To me it looks like there is a problem with paging activity. Is there a pager module running? Regards, --Naresh Narang ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/
Re: [ilugd] New machine, minor irritants
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Tue, Sep 9, 2008 at 12:30 AM, Raj Mathur wrote: Nah, CPUs, like I said, are at 55C +/-5C (depending on how many bzip2's I'm running to test the temperatures with :) . Sys is something else... You can try the phoronix test suite too. Best A. Mani -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.6 (GNU/Linux) Comment: http://getfiregpg.org iD8DBQFIxdVtoIK4BlImohYRAuR2AJ4/Ay+5ZPEyRH+U/ir7TC56ovYt4QCfU3aI Fbfshp+oHhaHcKk0+cIOQVw= =exW7 -END PGP SIGNATURE- -- A. Mani Member, Cal. Math. Soc ___ ilugd mailinglist -- ilugd@lists.linux-delhi.org http://frodo.hserus.net/mailman/listinfo/ilugd Archives at: http://news.gmane.org/gmane.user-groups.linux.delhi http://www.mail-archive.com/ilugd@lists.linux-delhi.org/