[opensuse] Swap partition "full"....
openSUSE 10.2 on a computer with 2GB of RAM and a 4GB swap partition Over time I've seen my swap partition get "full". It is currently sitting at 4029 of 4095 Mb with no applications running (outside of the KDE desktop). Things are fine until I do something that pushes things out of RAM and into swap... then the computer slows right down as it twiddles with whatever debris is left laying around in the swap and finds room to swap out new stuff. Is this normal? I have never noticed the swap partition filling up like this, and remaining "full" over an extended period including through logout and login (not reboot). Usually, when things get swapped to disk I see the partition use increase... then when it is no longer needed that swapped data is released/removed, and the swap partition falls back to zero. Lately, over the past 2 weeks (with no updates), that swap space is not being released. I know (expect) if I reboot, that the swap will reset to zero... or if I do a swapoff and then swapon I should see it all reset back to zero... but, that is not really telling me why the swap space is not released. Anyone have any insight on this? Or know a good resource on finding out what is stuck in the swap space? C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Thursday 08 November 2007 05:05:55 pm Clayton wrote:
openSUSE 10.2 on a computer with 2GB of RAM and a 4GB swap partition
Over time I've seen my swap partition get "full". It is currently sitting at 4029 of 4095 Mb with no applications running (outside of the KDE desktop).
Things are fine until I do something that pushes things out of RAM and into swap... then the computer slows right down as it twiddles with whatever debris is left laying around in the swap and finds room to swap out new stuff.
Is this normal? I have never noticed the swap partition filling up like this, and remaining "full" over an extended period including through logout and login (not reboot). Usually, when things get swapped to disk I see the partition use increase... then when it is no longer needed that swapped data is released/removed, and the swap partition falls back to zero. Lately, over the past 2 weeks (with no updates), that swap space is not being released.
I know (expect) if I reboot, that the swap will reset to zero... or if I do a swapoff and then swapon I should see it all reset back to zero... but, that is not really telling me why the swap space is not released.
Anyone have any insight on this? Or know a good resource on finding out what is stuck in the swap space?
top [f],[p],[Space] then [F],[p],[Space] Mind the case of the "f" You'll get the process list with the most swap at the top. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Anyone have any insight on this? Or know a good resource on finding out what is stuck in the swap space?
top [f],[p],[Space] then [F],[p],[Space]
Mind the case of the "f"
You'll get the process list with the most swap at the top.
Hmm that was quite interesting. I discovered a couple leftovers hanging around from a failed experiment with Joost. Cleaned them out. Swap use dropped a little... but it almost immediately went to the max available swap swapon -s Filename Type Size Used Priority /dev/hda2 partition 4192956 4192956 -1 These are the top swap using processes now: 145m mythbackend 142m ekiga 121m mysqld-max 116m firefox-bin 110m nscd 85m amarokapp 77m mythfrontend 55m pidgin 53m kmail 53m httpd2-prefork Nothing too extreme... there is more of course, but nothing that adds up to anything near the total physical RAM I have let alone the swap. I can't do a swapoff either. swapoff /dev/hda2 swapoff: /dev/hda2: Cannot allocate memory which I assume is because swap used is greater than available RAM, and the swapoff attempts to move whatever is on disk back to RAM? As soon as I start anything that tops off the available RAM space (launch a VMWare session for example), performance plummets... which makes sense since there is no swap partition available. I know.. I could just reboot... I probably will. It just seems so wrong that this is the only solution I can find to freeing up the lost swap space. C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Nov 8, 2007 10:17 AM, Clayton <smaug42@gmail.com> wrote:
145m mythbackend 142m ekiga 121m mysqld-max 116m firefox-bin 110m nscd 85m amarokapp 77m mythfrontend 55m pidgin 53m kmail 53m httpd2-prefork
Shut down mythbackend and frontend. IMHO mythtv allocates a lot of buffers, etc. At least it is not reboot, and you will remove one of the suspects :) -- Svetoslav Milenov (Sunny) Even the most advanced equipment in the hands of the ignorant is just a pile of scrap. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Shut down mythbackend and frontend. IMHO mythtv allocates a lot of buffers, etc. At least it is not reboot, and you will remove one of the suspects :)
Ok.. tried that. I killed off almost everything. Shut down apache2, mysql, mythtv, even logged out and went to init 3. Still swap is "full", and I am having the performance issues. Oh well... a reboot can't hurt... it's been a couple months.. :-P I just hate rebooting... it always messed up my mount points... for some reason I've never bothered to figure out, a couple of my drives swap around on every reboot... /dev/sde1 and /dev/sdd1 swich around and it screws everything up (apache etc) until a remount things where they are supposed to be mounted. C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Nov 8, 2007 11:03 AM, Clayton <smaug42@gmail.com> wrote:
I just hate rebooting... it always messed up my mount points... for some reason I've never bothered to figure out, a couple of my drives swap around on every reboot... /dev/sde1 and /dev/sdd1 swich around and it screws everything up (apache etc) until a remount things where they are supposed to be mounted.
Go in partitioner, and in advanced (I think) you can select to change the mount points by Label or UUID, that way they will not get messed. -- Svetoslav Milenov (Sunny) Even the most advanced equipment in the hands of the ignorant is just a pile of scrap. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I just hate rebooting... it always messed up my mount points... for some reason I've never bothered to figure out, a couple of my drives swap around on every reboot... /dev/sde1 and /dev/sdd1 swich around and it screws everything up (apache etc) until a remount things where they are supposed to be mounted.
Go in partitioner, and in advanced (I think) you can select to change the mount points by Label or UUID, that way they will not get messed.
It is set to device name right now for all drives. Strange that only sde1 and sdd1 get switched around (since all drives are mounted with the same settings)... they swap places with each boot. I reboot so rarely that it isn't a real issue. Hopefully the next reboot will be when I install 10.3. I have other stupid reboot issues... no sound when I reboot. i ahve to open YaST and Delete, and then Add the sound card back in. When I do that, I lose my desktop and have to log out and back in again. All stupid things I've never tracked down since they only pop up on a system restart. Anyway, rebooted, fixed the sound card, and mount points, logged out, and back in to fix the desktop that disappears when I delete the sound card (in YaST). RAM use is at 1125 of 2027 Mb and swap is 0 of 4095 with all of the same apps running (including apache2, mythtv etc). C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Clayton wrote:
I just hate rebooting... it always messed up my mount points... for some reason I've never bothered to figure out, a couple of my drives swap around on every reboot... /dev/sde1 and /dev/sdd1 swich around and it screws everything up (apache etc) until a remount things where they are supposed to be mounted.
I had that problem too. On a production server I have two big external USB hard drives which I use for "hot" backups via rsync. But on each reboot, the drives would randomly change orders and mount on the wrong mount points. The solution for this is to use the UUID of the drive (which is unique to the drive, I think) rather than the device node. So in /etc/fstab, I used to have something like: /dev/sde1 /b reiserfs defaults 0 0 /dev/sdf1 /b1 ext3 acl,user_xattr 0 0 but of course I couldn't control which drive became sde and which drive became sdf. But I looked up the UUID of each filesystem, and now my fstab looks like this: UUID=8948fa53-af4f-4db2-b3f7-f256e33ad20e /b reiserfs defaults 0 0 UUID=98ce791e-93a7-4c84-9652-5d1a1a8c437d /b1 ext3 acl,user_xattr 0 0 and it works perfectly every time! To get the UUID, have the (currently assigned) device node (I'll use /dev/sde1 as an example) and as root do: (For ext3) tune2fs -l /dev/sde1 (For reiser - partition must be unmounted first) reiserfstune /dev/sde1 These printouts will display your filesystems' UUIDs, which you can then sub in as shown above. Hope this helps! Glen -- Glen Barney gb2@c5i.net -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
The solution for this is to use the UUID of the drive (which is unique to the drive, I think) rather than the device node.
snip
But I looked up the UUID of each filesystem, and now my fstab looks like this:
UUID=8948fa53-af4f-4db2-b3f7-f256e33ad20e /b reiserfs defaults 0 0 UUID=98ce791e-93a7-4c84-9652-5d1a1a8c437d /b1 ext3 acl,user_xattr 0 0
and it works perfectly every time!
To get the UUID, have the (currently assigned) device node (I'll use /dev/sde1 as an example) and as root do:
(For ext3) tune2fs -l /dev/sde1
(For reiser - partition must be unmounted first) reiserfstune /dev/sde1
These printouts will display your filesystems' UUIDs, which you can then sub in as shown above.
Cool. Thanks. I'm saving this one for future use :-) C -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Clayton wrote:
Shut down mythbackend and frontend. IMHO mythtv allocates a lot of buffers, etc. At least it is not reboot, and you will remove one of the suspects :)
Ok.. tried that. I killed off almost everything. Shut down apache2, mysql, mythtv, even logged out and went to init 3. Still swap is "full", and I am having the performance issues. Oh well... a reboot can't hurt... it's been a couple months.. :-P
I just hate rebooting... it always messed up my mount points... for some reason I've never bothered to figure out, a couple of my drives swap around on every reboot... /dev/sde1 and /dev/sdd1 swich around and it screws everything up (apache etc) until a remount things where they are supposed to be mounted.
then label the partitions, and mount those filesystems by label instead of device name. I used to be a mount-by-device name only kind of guy.... when my system was exclusively SCSI, and I could control their device name by setting their SCSI ID numbers, regardless of where they were physically connected. But once I put a couple of IDE drives in...I found them just too much of a pain in the neck to keep track of what disk was plugged in where on those occasions when I disconnected their ribbon cables... On my laptop, the partition labels are like this: root_suse10_1 _usr _opt _home _local _tmp _windows_c Since the labels indicte the mount point, confusion is nearly impossible. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Clayton wrote:
Anyone have any insight on this? Or know a good resource on finding out what is stuck in the swap space? top [f],[p],[Space] then [F],[p],[Space]
Mind the case of the "f"
You'll get the process list with the most swap at the top.
Hmm that was quite interesting. I discovered a couple leftovers hanging around from a failed experiment with Joost. Cleaned them out. Swap use dropped a little... but it almost immediately went to the max available swap
swapon -s Filename Type Size Used Priority /dev/hda2 partition 4192956 4192956 -1
These are the top swap using processes now:
145m mythbackend 142m ekiga 121m mysqld-max 116m firefox-bin 110m nscd 85m amarokapp 77m mythfrontend 55m pidgin 53m kmail 53m httpd2-prefork
Nothing too extreme... there is more of course, but nothing that adds up to anything near the total physical RAM I have let alone the swap.
I can't do a swapoff either.
swapoff /dev/hda2 swapoff: /dev/hda2: Cannot allocate memory
which I assume is because swap used is greater than available RAM, and the swapoff attempts to move whatever is on disk back to RAM?
As soon as I start anything that tops off the available RAM space (launch a VMWare session for example), performance plummets... which makes sense since there is no swap partition available.
I know.. I could just reboot... I probably will. It just seems so wrong that this is the only solution I can find to freeing up the lost swap space.
C.
last time I saw this famd was the culprit... also caused performance issues and some other odd behaviour... restarted famd and problem went away... However, I was doing some testing with some code which at the time did use rather a lot of memory and it did not always exit terribly cleanly, mainly because I killed it rather frequently ... - -- ============================================================================== I have always wished that my computer would be as easy to use as my telephone. My wish has come true. I no longer know how to use my telephone. Bjarne Stroustrup ============================================================================== -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFHM0/jasN0sSnLmgIRArB5AJ4z/W+/vgNT87VZ5ZqtwsvoI6JRjwCfZySa TyF/cddLqREBc7OF/DQ/tj0= =LlRa -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Thursday 2007-11-08 at 17:17 +0100, Clayton wrote:
Hmm that was quite interesting. I discovered a couple leftovers hanging around from a failed experiment with Joost. Cleaned them out. Swap use dropped a little... but it almost immediately went to the max available swap
swapon -s Filename Type Size Used Priority /dev/hda2 partition 4192956 4192956 -1
These are the top swap using processes now:
145m mythbackend 142m ekiga 121m mysqld-max 116m firefox-bin 110m nscd 85m amarokapp 77m mythfrontend 55m pidgin 53m kmail 53m httpd2-prefork
Nothing too extreme... there is more of course, but nothing that adds up to anything near the total physical RAM I have let alone the swap.
I can't do a swapoff either.
swapoff /dev/hda2 swapoff: /dev/hda2: Cannot allocate memory
which I assume is because swap used is greater than available RAM, and the swapoff attempts to move whatever is on disk back to RAM?
I think you have a process somewhere with a memory hole: allocating memory and forgetting to free it later. I don't know if it is possible for the process to die without freeing the memory :-? top [M] usually shows it. - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFHNkYttTMYHG2NR9URAtz7AJ9HkYaIfH1Ch2dn5PesRMOMtta8aACfRi4d K4mlrxMstT4xkF2qkE+cO7Q= =hJbk -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
I think you have a process somewhere with a memory hole: allocating memory and forgetting to free it later. I don't know if it is possible for the process to die without freeing the memory :-?
top [M]
usually shows it.
I'm beginning to think the same. I can't nail it down though. I rebooted 3 days ago. Since then swap partition usage is slowly edging upward. Immediately after a reboot, and with all the regular services and applications running and the machine sitting at idle, roughly 1.2Gb of the 2.0Gb of RAM was allocated/used, and swap was at 0 of 4Gb. As I used the machine for my "normal" desktop use... so using Firefox, Thunderbird, Opera, Seamonkey, Smart, Skype, Ekiga, X-Plane, etc, the swap partition is up to 670/4095. Slowly... over time it fills up, and is not being released. This is only recently... say in the last month or so. I can't nail it down to what I updated recently that is doing this. C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 The Sunday 2007-11-11 at 11:10 +0100, Clayton wrote:
I think you have a process somewhere with a memory hole: allocating memory and forgetting to free it later. I don't know if it is possible for the process to die without freeing the memory :-?
top [M]
usually shows it.
I'm beginning to think the same. I can't nail it down though. I rebooted 3 days ago. Since then swap partition usage is slowly edging upward. Immediately after a reboot, and with all the regular services and applications running and the machine sitting at idle, roughly 1.2Gb of the 2.0Gb of RAM was allocated/used, and swap was at 0 of 4Gb. As I used the machine for my "normal" desktop use... so using Firefox, Thunderbird, Opera, Seamonkey, Smart, Skype, Ekiga, X-Plane, etc, the swap partition is up to 670/4095. Slowly... over time it fills up, and is not being released.
This is only recently... say in the last month or so. I can't nail it down to what I updated recently that is doing this.
Ideas: - Undo the updates in reverse order (rpm --last can tell you which). - Try accounting software (ac): rcacct start. I believe it can help, but I don't know how. The command 'sa' gives process info. Then, there most some tool that navigates memory blocks telling who is the owner of what, but again, I don't know which tool. In "gnome system monitor", in the view menu I see "memory maps" in grey, and I don't know how to activate it - but that's just the thing [...] ah, it shows the map for the highlilght process only. Too bad. I know i have seen some other tool somewhere... :-? - -- Cheers, Carlos E. R. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Made with pgp4pine 1.76 iD8DBQFHN2b9tTMYHG2NR9URAvtJAJ9O/YcREaB0PGye+XS/jH4k96Gu4QCePF9H PznHIZZLaFO5AldRGEtBOaM= =vjUt -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, Clayton wrote:
I'm beginning to think the same. I can't nail it down though. I rebooted 3 days ago. Since then swap partition usage is slowly edging upward. Immediately after a reboot, and with all the regular services and applications running and the machine sitting at idle, roughly 1.2Gb of the 2.0Gb of RAM was allocated/used, and swap was at 0 of 4Gb. As I used the machine for my "normal" desktop use... so using Firefox, Thunderbird, Opera, Seamonkey, Smart, Skype, Ekiga, X-Plane, etc, the swap partition is up to 670/4095. Slowly... over time it fills up, and is not being released.
This is only recently... say in the last month or so. I can't nail it down to what I updated recently that is doing this.
Glad to see I am not the only one observing this weird swap behaviour. I filed a bug report about this, but did not yet get a reply: https://bugzilla.novell.com/show_bug.cgi?id=340143 Can you check the output of "free" and "vmstat 3"? Do you also have plenty of memory in "free" and the system still swaps as mad? If you do, please add your findings to the bug report above. Thanks! Bye, LenZ - -- - ------------------------------------------------------------------ Lenz Grimmer <lenz@grimmer.com> -o) [ICQ: 160767607 | Jabber: LenZGr@jabber.org] /\\ http://www.lenzg.org/ V_V -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFHQXATSVDhKrJykfIRAv3oAJ93bX+tuigo9zz7r+LSUAtlLYWACACfbf+f KksuDtx8e8oMnQUOPOhrF78= =0ZyH -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Glad to see I am not the only one observing this weird swap behaviour. I filed a bug report about this, but did not yet get a reply:
https://bugzilla.novell.com/show_bug.cgi?id=340143
Can you check the output of "free" and "vmstat 3"? Do you also have plenty of memory in "free" and the system still swaps as mad? If you do, please add your findings to the bug report above. Thanks!
What you describe in the bug report is exactly the same as what I see on my 10.2 install.... and it's something fairly recent (say in the last 6 to 8 weeks... from something that was changed in an update?). One thing you said is almost a perfect description of my problem: --- "The swapping behaviour is fine for 1-2 days, then all the sudden the VM decides to swap out everything, and keeping lots of memorycompletely unused!" --- I'm getting this behaviour while using normal desktop apps... no cron job doing backups etc. I usually have plenty of memory available... less than half allocated when the system starts swapping madly and slowing down to the point where a reboot is the only way to recover normal behaviour. I will look into free and vmstat tonight. C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
Can you check the output of "free" and "vmstat 3"? Do you also have plenty of memory in "free" and the system still swaps as mad? If you do, please add your findings to the bug report above. Thanks!
OK, computer started to swap like mad this evening.... lots of free RAM.... but it is still swapping to disk, and not freeing the swap space up either. I will add a comment to the bug report.... free total used free shared buffers cached Mem: 2076040 1182912 893128 0 41784 559888 -/+ buffers/cache: 581240 1494800 Swap: 4192956 1472652 2720304 and vmstat 3 procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 0 1472652 891320 42088 561832 0 1 12 7 12 20 9 3 86 2 1 0 1472652 890736 42208 562360 0 0 0 41 898 1106 4 2 92 2 0 0 1472652 889832 42208 562980 0 0 21 0 804 1186 6 4 89 0 2 0 1472652 889460 42368 563304 0 0 0 56 822 1190 5 2 91 2 2 0 1472652 889148 42368 563708 0 0 27 0 790 1185 3 1 96 0 0 0 1472652 888400 42464 564148 0 0 21 2167 782 1145 7 3 89 1 1 0 1472652 887824 42652 564716 0 0 21 67 765 1099 3 2 93 3 1 0 1472652 887404 42652 565204 0 0 0 0 773 1117 4 1 94 0 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, Clayton wrote:
OK, computer started to swap like mad this evening.... lots of free RAM.... but it is still swapping to disk, and not freeing the swap space up either. I will add a comment to the bug report....
free total used free shared buffers cached Mem: 2076040 1182912 893128 0 41784 559888 -/+ buffers/cache: 581240 1494800 Swap: 4192956 1472652 2720304
Yes, this looks familiar. Plenty of memory free, but still lots of swap used. However:
vmstat 3 procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- r b swpd free buff cache si so bi bo in cs us sy id wa 0 0 1472652 891320 42088 561832 0 1 12 7 12 20 9 3 86 2 1 0 1472652 890736 42208 562360 0 0 0 41 898 1106 4 2 92 2 0 0 1472652 889832 42208 562980 0 0 21 0 804 1186 6 4 89 0 2 0 1472652 889460 42368 563304 0 0 0 56 822 1190 5 2 91 2 2 0 1472652 889148 42368 563708 0 0 27 0 790 1185 3 1 96 0 0 0 1472652 888400 42464 564148 0 0 21 2167 782 1145 7 3 89 1 1 0 1472652 887824 42652 564716 0 0 21 67 765 1099 3 2 93 3 1 0 1472652 887404 42652 565204 0 0 0 0 773 1117 4 1 94 0
According to this, it's not actively swapping at all! See the "si/so" columns. So I wonder what's causing the disk activity? In any case, it looks fishy. Bye, LenZ - -- - ------------------------------------------------------------------ Lenz Grimmer <lenz@grimmer.com> -o) [ICQ: 160767607 | Jabber: LenZGr@jabber.org] /\\ http://www.lenzg.org/ V_V -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4-svn0 (GNU/Linux) Comment: Using GnuPG with SUSE - http://enigmail.mozdev.org iD8DBQFHQfGuSVDhKrJykfIRAgv2AJ9fAzktuHlEE2ujxb9fo4sXXNPUMQCeJVnf mCXFZQ6t+iHAyR14ZCYTX4Q= =C/TV -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Monday 19 November 2007 21:27:27 Lenz Grimmer wrote:
Hi,
Clayton wrote:
OK, computer started to swap like mad this evening.... lots of free RAM.... but it is still swapping to disk, and not freeing the swap space up either. I will add a comment to the bug report....
free total used free shared buffers cached Mem: 2076040 1182912 893128 0 41784 559888 -/+ buffers/cache: 581240 1494800 Swap: 4192956 1472652 2720304
Yes, this looks familiar. Plenty of memory free, but still lots of swap
That's usually a sign that something big was running, and eating up memory, forcing other things into swap, and then it exited (or crashed), freeing up the memory it used. Seeing 50% free memory on a linux system that has been running for a while is highly unusual. It should be used for buffers and cache, if for nothing else. It is almost always a sign of the above scenario Anders -- Madness takes its toll -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
free total used free shared buffers cached Mem: 2076040 1182912 893128 0 41784 559888 -/+ buffers/cache: 581240 1494800 Swap: 4192956 1472652 2720304
Yes, this looks familiar. Plenty of memory free, but still lots of swap
That's usually a sign that something big was running, and eating up memory, forcing other things into swap, and then it exited (or crashed), freeing up the memory it used.
Seeing 50% free memory on a linux system that has been running for a while is highly unusual. It should be used for buffers and cache, if for nothing else. It is almost always a sign of the above scenario
True enough. I had been running a few things (Acroread being one) that had been closed just moments before I ran free. Same goes for vmstat... I ran it after things had settled down, not when swap was being loaded up (wasn't thinking...). When the apps were closed, the RAM was freed up (as expected). It is worth noting though, that the 1440Mb of swap has been there for 2 days now. RAM has ranged up and down as I use apps... on a clean boot with only the background fluff running plus KDE and the apps I run on startup there, I see around 1200Mb of RAM allocated (which then fairly rapidly tops out the RAM with buffers and cache stuff). Normal behavior on my setup which has been running the same apps since 10.2 was released was to rarely if ever put anything into swap. Swap would sit at zero for weeks... or longer. I will try to run free and vmstat when things are going a little nuts and get a better picture of memory and swap usage.... the annoying thing about this problem is that sometimes it works perfectly as expected. I just ran a test... starting at about 1200Mb RAM and 1400Mb swap, I launched X-Plane, Firefox, Opera, Seamonkey, VMWare (running 10.2), and VirtualBox (running WinXP). RAM maxed out and things started swapping with swap going from 1400Mb to 2200Mb - what I would call expected normal behaviour. Exited all the applications, and RAM was released, and some swap was released and it's now sitting at 1572Mb. If I close down everything, log out and log back in, swap usage doesn't move.. or if anything swap use increases slightly on login.now it will never drop below 1572Mb swap until I reboot. I expect that it'll top over 2000Mb in the next 24 hours... most of that while sitting idle. Now the annoying thing about this prob is.... sometimes for no easily discernible reason, during normal use (not the stress test I did above) RAM will not be maxed out, and the system will start paging out to swap... and swap use will go up.. and not be released. eventually swap is maxed out, and the computer basically becomes unusable. Reboot fixes. I am nowhere near an expert on this.. only a vague understanding.. more like a peripheral awareness of the magic going on in this part of my computer.... so... C. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Monday 19 November 2007 22:22:54 Clayton wrote:
free total used free shared buffers cached Mem: 2076040 1182912 893128 0 41784 559888 -/+ buffers/cache: 581240 1494800 Swap: 4192956 1472652 2720304
Yes, this looks familiar. Plenty of memory free, but still lots of swap
That's usually a sign that something big was running, and eating up memory, forcing other things into swap, and then it exited (or crashed), freeing up the memory it used.
Seeing 50% free memory on a linux system that has been running for a while is highly unusual. It should be used for buffers and cache, if for nothing else. It is almost always a sign of the above scenario
True enough. I had been running a few things (Acroread being one) that had been closed just moments before I ran free. Same goes for vmstat... I ran it after things had settled down, not when swap was being loaded up (wasn't thinking...). When the apps were closed, the RAM was freed up (as expected). It is worth noting though, that the 1440Mb of swap has been there for 2 days now. RAM has ranged up and down as I use apps... on a clean boot with only the background fluff running plus KDE and the apps I run on startup there, I see around 1200Mb of RAM allocated (which then fairly rapidly tops out the RAM with buffers and cache stuff). Normal behavior on my setup which has been running the same apps since 10.2 was released was to rarely if ever put anything into swap. Swap would sit at zero for weeks... or longer.
What does /proc/meminfo look like on that machine when this is happening? Anders -- Madness takes its toll -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
On Mon, 2007-11-19 at 22:22 +0100, Clayton wrote: <SNIP>
I will try to run free and vmstat when things are going a little nuts and get a better picture of memory and swap usage.... the annoying thing about this problem is that sometimes it works perfectly as expected. I just ran a test... starting at about 1200Mb RAM and 1400Mb swap, I launched X-Plane, Firefox, Opera, Seamonkey, VMWare (running 10.2), and VirtualBox (running WinXP). RAM maxed out and things started swapping with swap going from 1400Mb to 2200Mb - what I would call expected normal behaviour. Exited all the applications, and RAM was released, and some swap was released and it's now sitting at 1572Mb. If I close down everything, log out and log back in, swap usage doesn't move.. or if anything swap use increases slightly on login.now it will never drop below 1572Mb swap until I reboot. I expect that it'll top over 2000Mb in the next 24 hours... most of that while sitting idle.
Now the annoying thing about this prob is.... sometimes for no easily discernible reason, during normal use (not the stress test I did above) RAM will not be maxed out, and the system will start paging out to swap... and swap use will go up.. and not be released. eventually swap is maxed out, and the computer basically becomes unusable. Reboot fixes.
Well you did not mention SuSE 9.2 and Evolution so perhaps its one of the services root loads when getting to run level 5. Try two things, first <crtl> <alt> F1 and as root init 3 and let it run about ten minutes then check top as suggested. If its still a problem do init 1 and top as before. Also try this over night reboot to run level 1,3 and 5 on separate nights and see where swap gets to by checking in C-A-F1 If either of these cures the problem its something root or the system puts up. ___ _ _ _ ____ _ _ _ | | | | [__ | | | |___ |_|_| ___] | \/ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org
participants (10)
-
Aaron Kulkis -
Anders Johansson -
Carl Spitzer -
Carlos E. R. -
Clayton -
G T Smith -
Glen -
Lenz Grimmer -
Silviu Marin-Caea -
Sunny