[opensuse] Is this normal? systemd-journal taking nearly 2 gigs of memory.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Top: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal 1853m? That much? It is virtual ram, not actual ram, but still... Can I restart it? Should I? Of course, it is not the only hungry app. These are the worst: 5733 cer 20 0 4944m 2,4g 18m S 45,8 30,2 2041:15 firefox 21392 cer 20 0 1421m 541m 519m S 5,3 6,8 19:39.09 vmware-vmx 2765 vscan 20 0 714m 246m 1024 S 0,0 3,1 2:16.10 clamd 21246 cer 20 0 1383m 191m 16m S 0,0 2,4 94:17.99 thunderbird-bin 4338 root 20 0 318m 55m 10m S 6,3 0,7 265:38.72 Xorg 5984 cer 20 0 4432m 51m 3184 S 0,0 0,6 3:16.25 java 21163 cer 20 0 193m 44m 3264 S 0,0 0,6 3:26.20 pine 318 root 20 0 1853m 41m 40m S 0,0 0,5 3:41.14 systemd-journal It ends in me having 3 GB of swap in use on a machine with 8 GiB of RAM. - -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) iEYEARECAAYFAlLXy7YACgkQtTMYHG2NR9XalgCeKf2qujCqwk8NcQCSU/ZULfmA Sj8AnRQVRwXukLRjReBTXfKef/Q5Mi3Y =oNXU -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 16/01/14 12:08, Carlos E. R. wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal
1853m? That much? It is virtual ram, not actual ram, but still...
AIUI, the VIRT figure includes the on-disk size of the binary and all its linked libraries, as well as the resident data and (parts of) binaries. It certainly has no relation to any actual memory or disk usage (not least because many libraries will be counted in several applications' figure...)
Can I restart it? Should I?
Of course, it is not the only hungry app. These are the worst:
5733 cer 20 0 4944m 2,4g 18m S 45,8 30,2 2041:15 firefox 21392 cer 20 0 1421m 541m 519m S 5,3 6,8 19:39.09 vmware-vmx 2765 vscan 20 0 714m 246m 1024 S 0,0 3,1 2:16.10 clamd 21246 cer 20 0 1383m 191m 16m S 0,0 2,4 94:17.99 thunderbird-bin 4338 root 20 0 318m 55m 10m S 6,3 0,7 265:38.72 Xorg 5984 cer 20 0 4432m 51m 3184 S 0,0 0,6 3:16.25 java 21163 cer 20 0 193m 44m 3264 S 0,0 0,6 3:26.20 pine 318 root 20 0 1853m 41m 40m S 0,0 0,5 3:41.14 systemd-journal
It ends in me having 3 GB of swap in use on a machine with 8 GiB of RAM.
- -- Cheers / Saludos,
Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux)
iEYEARECAAYFAlLXy7YACgkQtTMYHG2NR9XalgCeKf2qujCqwk8NcQCSU/ZULfmA Sj8AnRQVRwXukLRjReBTXfKef/Q5Mi3Y =oNXU -----END PGP SIGNATURE-----
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Le 16/01/2014 13:08, Carlos E. R. a écrit :
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal
don't know what that "virt" column mean what do "free" gives? free total used free shared buffers cached Mem: 8115060 2472104 5642956 22368 108348 1031216 -/+ buffers/cache: 1332540 6782520 Swap: 2109436 0 2109436 (I also have big firefox in top) http://mugurel.sumanariu.ro/linux/the-difference-among-virt-res-and-shr-in-t... jdd -- http://www.dodin.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
jdd said the following on 01/16/2014 08:01 AM:
Le 16/01/2014 13:08, Carlos E. R. a écrit :
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal
don't know what that "virt" column mean
In sort, it is the address space allocated to that process. It has nothing to do with the amount of memory the process is using, which is the RESident set. A well designed application will make heavy use of SHaRed libraries rather than replicating the code itself. Nartually some applications will be very specialized and simply can't share some critical code. GUI things like GIMP, OpenOffice and Darktable might be examples. Other groups of applications may share data libraries (icon sets for example).
what do "free" gives?
free total used free shared buffers cached Mem: 8115060 2472104 5642956 22368 108348 1031216 -/+ buffers/cache: 1332540 6782520 Swap: 2109436 0 2109436
Well you are not swapping or making use of swap to save page-out. That's good. You are running entirely in memory. PERHAPS you are CPU limited. At the top of 'top' there is the 'load average' and you can get a report on where you CPU is spending its time. Your version of 'top' should allow you to see that for each of your CPU cores. Is systemd-journal taking nearly 2 gigs of memory? No! It is taking 43Meg of resident space. Look to the virtual address space of other processes! The whole point of VIRTUAL memory is just that; its only 'virtually' there. Go google for some background on virtual memory. Its a concept from the 1950s -- most of what we have in computing was thought of and tried by then but they lacked the technology to use it all effectively. As far as the innards of the application go. how it handles file mapping, its internal caching, all that is another matter. It may be of interest in fine tuning, but any sensible metric will first ask "Is this application a limiting one?" before tuning. -- How long did the whining go on when KDE2 went on KDE3? The only universal constant is change. If a species can not adapt it goes extinct. That's the law of the universe, adapt or die. -- Billie Walsh, May 18 2013 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Content-ID:
Le 16/01/2014 13:08, Carlos E. R. a écrit :
Top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal
don't know what that "virt" column mean
what do "free" gives?
free total used free shared buffers cached Mem: 8115060 2472104 5642956 22368 108348 1031216 -/+ buffers/cache: 1332540 6782520 Swap: 2109436 0 2109436
Telcontar:~ # free -h total used free shared buffers cached Mem: 7.8G 4.2G 3.6G 0B 122M 2.1G - -/+ buffers/cache: 1.9G 5.9G Swap: 20G 1.5G 19G Telcontar:~ #
(I also have big firefox in top)
I restarted it, now the figures are 4194 cer 20 0 1575m 701m 58m S 4,0 8,8 4:31.05 firefox it was: 5733 cer 20 0 4944m 2,4g 18m S 45,8 30,2 2041:15 firefox
http://mugurel.sumanariu.ro/linux/the-difference-among-virt-res-and-shr-in-t...
Mmm. - -- Cheers, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) iEYEARECAAYFAlLX54YACgkQtTMYHG2NR9Ur3ACdFyfKt6gPlm5/hgNwMGAQe35F exUAniMyymNcqAAzR0mOpR3CYTbVN2m7 =/OmZ -----END PGP SIGNATURE-----
El 16/01/14 09:08, Carlos E. R. escribió:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Top:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND ... 318 root 20 0 1853m 43m 43m S 0,0 0,5 3:40.99 systemd-journal
1853m? That much? It is virtual ram, not actual ram, but still...
Read the following thread http://comments.gmane.org/gmane.comp.sysutils.systemd.devel/11362 To determine how much memory is the process really using run pmap `pidof systemd-journald` see the "total" at the end in the PSS column. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Thursday, 2014-01-16 at 10:47 -0300, Cristian Rodríguez wrote:
El 16/01/14 09:08, Carlos E. R. escribió:
1853m? That much? It is virtual ram, not actual ram, but still...
Read the following thread
http://comments.gmane.org/gmane.comp.sysutils.systemd.devel/11362
Interesting.
To determine how much memory is the process really using run
pmap `pidof systemd-journald`
see the "total" at the end in the PSS column.
Large dump. But you say only the last part is interesting. 318: systemd-journal START SIZE RSS PSS DIRTY SWAP PERM MAPPING 0000000000400000 172K 124K 124K 0K 0K r-xp /usr/lib/systemd/systemd-journald 000000000062a000 4K 4K 4K 0K 0K r--p /usr/lib/systemd/systemd-journald 000000000062b000 4K 0K 0K 0K 4K rw-p /usr/lib/systemd/systemd-journald 0000000001be4000 236K 124K 124K 124K 44K rw-p [heap] 00007f7bf0cfb000 4104K 28K 7K 28K 0K rw-s /run/log/journal/2ce1d54548517a7307c1c2bc38206d00/system.journal 00007f7bf10fd000 8192K 1272K 821K 1272K 0K rw-s /run/log/journal/2ce1d54548517a7307c1c2bc38206d00/system.journal ... 00007fff929ef000 4K 4K 0K 0K 0K r-xp [vdso] ffffffffff600000 4K 0K 0K 0K 0K r-xp [vsyscall] Total: 1988064K 44360K 15251K 43284K 224K 1076K writable-private, 38076K readonly-private, 1948912K shared, and 43488K referenced I'm not familiar with this kind of output. The VIRT figure of top must be the "SIZE" figure here. It is growing (1941m/1988064K this instant). - -- Cheers, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar) -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (GNU/Linux) iEYEARECAAYFAlLX6yUACgkQtTMYHG2NR9VNaQCbBU0YRnTwDtGfy+98bKVZNVUE kugAoJXDxzpu6de8S6SAVJiJ9I4FKHEL =XD2g -----END PGP SIGNATURE-----
El 16/01/14 11:22, Carlos E. R. escribió:
00007fff929ef000 4K 4K 0K 0K 0K r-xp [vdso] ffffffffff600000 4K 0K 0K 0K 0K r-xp [vsyscall] Total: 1988064K 44360K 15251K 43284K 224K
See ..the journal is using ~14.8 Megabytes of memory. not gigabytes of it. the rest is the process address space (NOT "used" memory) plus memory that is shared with other process of the system (shared libraries mostly or totally) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 15:31, Cristian Rodríguez wrote:
El 16/01/14 11:22, Carlos E. R. escribió:
00007fff929ef000 4K 4K 0K 0K 0K r-xp [vdso] ffffffffff600000 4K 0K 0K 0K 0K r-xp [vsyscall] Total: 1988064K 44360K 15251K 43284K 224K
See ..the journal is using ~14.8 Megabytes of memory. not gigabytes of it. the rest is the process address space (NOT "used" memory) plus memory that is shared with other process of the system (shared libraries mostly or totally)
Ok, good. Then the next question is, how do I find what applications are using swap? There is about 1.5 GB in use. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Carlos E. R. said the following on 01/16/2014 09:38 AM:
Then the next question is, how do I find what applications are using swap? There is about 1.5 GB in use.
'Top' can be configured to show swap. RTFM or use the help -- '?' function -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 15:49, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 09:38 AM:
Then the next question is, how do I find what applications are using swap? There is about 1.5 GB in use.
'Top' can be configured to show swap. RTFM or use the help -- '?' function
I did that many times. I know how to display swap usage, but not how to sort on it. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
El 16/01/14 11:38, Carlos E. R. escribió:
On 2014-01-16 15:31, Cristian Rodríguez wrote:
El 16/01/14 11:22, Carlos E. R. escribió:
00007fff929ef000 4K 4K 0K 0K 0K r-xp [vdso] ffffffffff600000 4K 0K 0K 0K 0K r-xp [vsyscall] Total: 1988064K 44360K 15251K 43284K 224K
See ..the journal is using ~14.8 Megabytes of memory. not gigabytes of it. the rest is the process address space (NOT "used" memory) plus memory that is shared with other process of the system (shared libraries mostly or totally)
Ok, good.
Then the next question is, how do I find what applications are using swap? There is about 1.5 GB in use.
Try this #! /bin/bash # # swap.sh: Shows the swap usage of each process # Author: Robert Love swap_total=0 for i in /proc/[0-9]*; do pid=$(echo $i | sed -e 's/\/proc\///g') swap_pid=$(cat /proc/$pid/smaps | awk 'BEGIN{total=0}/^Swap:/{total+=$2}END{print total}') if [ "$swap_pid" -gt 0 ]; then name=$(cat /proc/$pid/status | grep ^Name: | awk '{print $2}') echo "${name} (${pid}) ${swap_pid} kB" let swap_total+=$swap_pid fi done echo echo "Total: ${swap_total} kB" This will not give you precise results, but will probably help you to figure out things. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 16:04, Cristian Rodríguez wrote:
El 16/01/14 11:38, Carlos E. R. escribió:
This will not give you precise results, but will probably help you to figure out things.
A bit, thanks... it is not sorted. I see many things using thousands of kilobytes, but that amounts to only a few megabytes. I'll try adding a sort to that output. [...] A crude sort, meanwhile. # echo "${name} (${pid}) ${swap_pid} kB" echo "${swap_pid} kB ${name} (${pid}) " then run: swapusage | sort -g ... 10472 kB polkitd (4541) 12104 kB oyranos-monitor (5362) 12416 kB applet.py (5355) 13988 kB Xorg (4338) 14504 kB spamd (2006) 14544 kB spamd (1992) 15420 kB wish (5394) 19276 kB vmplayer (3778) 22988 kB /usr/sbin/amavi (2218) 23112 kB /usr/sbin/amavi (2362) 23124 kB /usr/sbin/amavi (2353) 23352 kB scp-dbus-servic (21923) 23712 kB /usr/sbin/amavi (2167) 23792 kB /usr/sbin/amavi (2418) 30140 kB pine (21163) 41132 kB named (2767) 43856 kB /usr/sbin/amavi (3786) 53296 kB /usr/sbin/spamd (30213) 64228 kB mysqld (3127) 122156 kB java (5984) 168756 kB thunderbird-bin (21246) Total: 1293920 kB Nothing really big, but they add. The worst, java and thunderbird. Hey, look at pine! Sigh... I guess that 8 GiB is not that much nowdays. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
On Thu, Jan 16, 2014 at 10:43 AM, Carlos E. R.
Total: 1293920 kB
Nothing really big, but they add. The worst, java and thunderbird. Hey, look at pine!
Sigh... I guess that 8 GiB is not that much nowdays.
That's 1.29 GB, not 12. Your 8GB still seems like plenty. Greg -- Greg Freemyer -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 17:01, Greg Freemyer wrote:
On Thu, Jan 16, 2014 at 10:43 AM, Carlos E. R. <> wrote:
Sigh... I guess that 8 GiB is not that much nowdays.
That's 1.29 GB, not 12. Your 8GB still seems like plenty.
I never said 12. Where? :-? I have 8 GiB ram, with 1.5G of used swap. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
On Thu, Jan 16, 2014 at 11:32 AM, Carlos E. R.
On 2014-01-16 17:01, Greg Freemyer wrote:
On Thu, Jan 16, 2014 at 10:43 AM, Carlos E. R. <> wrote:
Sigh... I guess that 8 GiB is not that much nowdays.
That's 1.29 GB, not 12. Your 8GB still seems like plenty.
I never said 12. Where? :-?
I have 8 GiB ram, with 1.5G of used swap.
I guess you confused me when you basically said: 1.3 GB of ram used by apps, I guess 8GB is not that much nowdays. I don't know what is driving your RAM usage, but it doesn't seem to apps. Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 18:00, Greg Freemyer wrote:
On Thu, Jan 16, 2014 at 11:32 AM, Carlos E. R. <> wrote:
I guess you confused me when you basically said:
1.3 GB of ram used by apps, I guess 8GB is not that much nowdays.
I don't know what is driving your RAM usage, but it doesn't seem to apps.
Ah. Well, yes, I meant that 8 GiB of RAM in chips is not that much nowdays, if my system has 1.5 in swap. There is 3,3G free and 2,4G in cache, the system is working fine; actually, having 1.5 in swap allows me to have that big cache and free pools, otherwise they'd be smaller. Maybe I have that much in swap simply because I hibernate the system every night. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Maybe I have that much in swap simply because I hibernate the system every night.
I just hibernated my laptop. 4MB of swap in use prior to hibernate. 60 MB immediately after hibernate/wakeup. You'll need to find a different reason it looks like. Greg -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 19:08, Greg Freemyer wrote:
Maybe I have that much in swap simply because I hibernate the system every night.
I just hibernated my laptop.
4MB of swap in use prior to hibernate. 60 MB immediately after hibernate/wakeup.
You'll need to find a different reason it looks like.
And some heavy memory users. This is the current list, sorted by swap use: 10416 kB polkitd (4541) 12104 kB oyranos-monitor (5362) 12388 kB applet.py (5355) 13584 kB Xorg (4338) 14504 kB spamd (2006) 14544 kB spamd (1992) 15420 kB wish (5394) 18952 kB vmplayer (3778) 22972 kB /usr/sbin/amavi (2218) 23000 kB /usr/sbin/amavi (2353) 23000 kB /usr/sbin/amavi (2362) 23352 kB scp-dbus-servic (21923) 23636 kB /usr/sbin/amavi (2418) 23688 kB /usr/sbin/amavi (2167) 30140 kB pine (21163) 41108 kB named (2767) 43856 kB /usr/sbin/amavi (3786) 53296 kB /usr/sbin/spamd (30213) 64228 kB mysqld (3127) 121264 kB java (5984) 167080 kB thunderbird-bin (21246) -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Carlos E. R. said the following on 01/16/2014 12:59 PM:
Maybe I have that much in swap simply because I hibernate the system every night.
Very likely. Why don't you try turning swap off then on again. -- How long did the whining go on when KDE2 went on KDE3? The only universal constant is change. If a species can not adapt it goes extinct. That's the law of the universe, adapt or die. -- Billie Walsh, May 18 2013 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 19:42, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 12:59 PM:
Maybe I have that much in swap simply because I hibernate the system every night.
Very likely.
Why don't you try turning swap off then on again.
That's worse. Then memory that is currently free is used by what was out, in swap. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Carlos E. R. said the following on 01/16/2014 02:19 PM:
On 2014-01-16 19:42, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 12:59 PM:
Maybe I have that much in swap simply because I hibernate the system every night.
Very likely.
Why don't you try turning swap off then on again.
That's worse. Then memory that is currently free is used by what was out, in swap.
Possibly; possibly not. If the swap was allocated and used when in hibernate then its not in use by any process any more, but still 'allocated' as in 'high water mark'. Turning the swap device off de-allocates memory that isn't mapped for any process. Turn it back on and it sets the high water mark to zero. Right now you have lots of free memory. You have no reason for the current process set to be using swap. So long as there is memory available, either unused or reclaimable from the VM system, Linux won't use swap. If any process actually is using swap when you have free memory then something is very very wrong with your system! Very wrong! The VM system is, in effect, permanently in 'page out' mode, but using any page again bring it to the back of the queue. Only if there is no free memory will a page drop off the front of the queue "into swap". It its a code page then there is no reason to 'swap' it out, its already out there - in the executable or library file. If the data page gets references its brought to the back of the queue. "Only data pages ever get swapped out." Then there's IO buffers. But that's IO not swap. That you have free memory AT ALL means that there are pages on the free queue that have never been allocated at all. If the VM systems needs memory for a process and its not on the active queue, then it will take a page from the free queue and it goes on the active queue, which is LRU and whose front end might get swapped out. But so long as there is free memory there is no need to 'claim' a page from the active queue and push its contents to swap. Free memory means no swapping/paging. If you have free memory then why should there be anything in swap? There isn't, that's a 'high water mark' from when there was. And its from when you backed up physical memory when you hibernated. OK, I've skipped detail and used a very broad/generalized brush about how VM works, but the basic point holds: If You Have Free Memory Then You Aren't Using Swap. Modulo fringe cases. So try the swapoff/on and see what changes. http://www.redhat.com/magazine/001nov04/features/vm/#life-page https://www.kernel.org/doc/gorman/pdf/understand.pdf -- While wizards don't believe in gods they know for a fact that gods believe in gods. (Reaper Man) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 23:40, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 02:19 PM:
Why don't you try turning swap off then on again.
That's worse. Then memory that is currently free is used by what was out, in swap.
Possibly; possibly not.
Look: (using "quoted" text because it disables line wrap in Thunderbird)
Telcontar:~ # free -h total used free shared buffers cached Mem: 7.8G 5.7G 2.1G 0B 424M 3.1G -/+ buffers/cache: 2.2G 5.6G Swap: 20G 1.4G 19G Telcontar:~ # swapoff -a
(this took a minute or two, with about 50% cpu load)
Telcontar:~ # free -h total used free shared buffers cached Mem: 7.8G 6.7G 1.1G 0B 424M 3.0G -/+ buffers/cache: 3.3G 4.5G Swap: 0B 0B 0B Telcontar:~ # swapon -a Telcontar:~ #
My free memory went down by 1 GiB, cached just by 0.1 GiB. Used memory went up by 1 GiB, which matches. About 0.4 GiB of what was in swap was discarded, it seems. The end result is that disabling swap I get less free memory, as I said, although not as much as I thought :-)
Telcontar:~ # free -h total used free shared buffers cached Mem: 7.8G 6.7G 1.1G 0B 421M 3.0G -/+ buffers/cache: 3.3G 4.5G Swap: 20G 24K 20G Telcontar:~ #
-- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Carlos E. R. said the following on 01/16/2014 06:06 PM:
The end result is that disabling swap I get less free memory, as I said, although not as much as I thought:-)
Which raises some interesting questions: if you have ANY free memory, why is swap being used at all? In the limiting case, that you have over a Gig of free memory, then with your present process set you don't seem to need swap at all. The 'swapoff' worked. You could have left it off. Were any error logged during the time it was off? Are you using a tmpfs by any chance? -- How long did the whining go on when KDE2 went on KDE3? The only universal constant is change. If a species can not adapt it goes extinct. That's the law of the universe, adapt or die. -- Billie Walsh, May 18 2013 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-17 01:55, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 06:06 PM:
The end result is that disabling swap I get less free memory, as I said, although not as much as I thought:-)
Which raises some interesting questions: if you have ANY free memory, why is swap being used at all?
Because hibernation forces everything into swap. If some chunks are not needed when the computer is thawed, they stay out of the way in swap, leaving more memory free.
In the limiting case, that you have over a Gig of free memory, then with your present process set you don't seem to need swap at all. The 'swapoff' worked. You could have left it off. Were any error logged during the time it was off?
If I leave it off the computer is slower. Don't you see? And I use hibernation, I need it.
Are you using a tmpfs by any chance?
No. Well, yes, of course, those that systemd requires:
Telcontar:~ # df -h | grep tmpfs
Filesystem Size Used Avail Use% Mounted on
devtmpfs 3.9G 100K 3.9G 1% /dev tmpfs 4.0G 1.1M 4.0G 1% /dev/shm tmpfs 4.0G 407M 3.6G 11% /run tmpfs 4.0G 0 4.0G 0% /sys/fs/cgroup tmpfs 4.0G 407M 3.6G 11% /var/lock tmpfs 4.0G 407M 3.6G 11% Telcontar:~ #
/run, /var/lock and /var/run are the same one. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
Carlos E. R. said the following on 01/16/2014 10:05 PM:
In the limiting case, that you have over a Gig of free memory, then with your present process set you don't seem to need swap at all. The 'swapoff' worked. You could have left it off. Were any error logged during the time it was off?
If I leave it off the computer is slower. Don't you see?
Explain please.
And I use hibernation, I need it.
Yes, I understand that. So why are you complaining? You and John make a good case for its inevitability. -- How long did the whining go on when KDE2 went on KDE3? The only universal constant is change. If a species can not adapt it goes extinct. That's the law of the universe, adapt or die. -- Billie Walsh, May 18 2013 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Carlos E. R. said the following on 01/16/2014 10:05 PM:
Are you using a tmpfs by any chance? No. Well, yes, of course, those that systemd requires:
Telcontar:~ # df -h | grep tmpfs
Filesystem Size Used Avail Use% Mounted on devtmpfs 3.9G 100K 3.9G 1% /dev tmpfs 4.0G 1.1M 4.0G 1% /dev/shm tmpfs 4.0G 407M 3.6G 11% /run tmpfs 4.0G 0 4.0G 0% /sys/fs/cgroup tmpfs 4.0G 407M 3.6G 11% /var/lock tmpfs 4.0G 407M 3.6G 11% Telcontar:~ #
What are those 'swap' numbers again? -- How long did the whining go on when KDE2 went on KDE3? The only universal constant is change. If a species can not adapt it goes extinct. That's the law of the universe, adapt or die. -- Billie Walsh, May 18 2013 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
Carlos E. R. said the following on 01/16/2014 10:43 AM:
Total: 1293920 kB
How does that compare to swapon -s and the top few lines of 'top' I'm just wondering about the consistency of results using different methods. I have top -b -n 1 | head -6 top - 11:10:07 up 1 day, 1:28, 8 users, load average: 0.63, 0.83, 0.94 Tasks: 155 total, 1 running, 154 sleeping, 0 stopped, 0 zombie %Cpu: 24.2 us, 3.7 sy, 0.1 ni, 69.5 id, 2.4 wa, 0.0 hi, 0.2 si, 0.0 st KiB Mem: 4054448 total, 3813068 used, 241380 free, 8 buffers KiB Swap: 1468412 total, 50192 used, 1418220 free, 1705252 cached and /usr/sbin/swapon -s Filename Type Size Used Priority /dev/sda1 partition 1468412 50192 -1 4G would be plenty if it wasn't for all those firefox tabs :-/ A 4 or 8 core CPU would be nice even with only 4G of memory. My laptop, with only 1.5G and a slower CPU is more responsive because of the additional cores. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
On 2014-01-16 17:15, Anton Aylward wrote:
Carlos E. R. said the following on 01/16/2014 10:43 AM:
Total: 1293920 kB
How does that compare to
swapon -s
and the top few lines of 'top'
I'm just wondering about the consistency of results using different methods.
Mmmm. I modified the code to have it sorted, but now the total is not working: swap_total=0 function bucle() { for i in /proc/[0-9]*; do pid=$(echo $i | sed -e 's/\/proc\///g') swap_pid=$(cat /proc/$pid/smaps | awk 'BEGIN{total=0}/^Swap:/{total+=$2}END{print total}') if [ "$swap_pid" -gt 0 ]; then name=$(cat /proc/$pid/status | grep ^Name: | awk '{print $2}') # echo "${swap_pid} kB ${name} (${pid}) " printf "%12i kB %30s (%i)\n" ${swap_pid} ${name} ${pid} let swap_total+=$swap_pid fi done } bucle | sort -g echo echo "Total: ${swap_total} kB" I now get: Total: 0 kB so I must have a mistake in my code above, the "swap_total" var is not treated as a global var inside the function. So I can not currently compare all methods.
4G would be plenty if it wasn't for all those firefox tabs :-/
Indeed....
A 4 or 8 core CPU would be nice even with only 4G of memory. My laptop, with only 1.5G and a slower CPU is more responsive because of the additional cores.
You need everything, cores and ram. -- Cheers / Saludos, Carlos E. R. (from 12.3 x86_64 "Dartmouth" at Telcontar)
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 01/16/2014 06:08 AM, Carlos E. R. wrote:
It ends in me having 3 GB of swap in use on a machine with 8 GiB of RAM.
/etc/systemd/journald.conf SystemMaxUse=320M - -- David C. Rankin, J.D.,P.E. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.16 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlLYtBYACgkQZMpuZ8Cyrcg5tgCeMHq02cQADCzLRJaKW6wUFub2 lKcAn2nZM/FewK245vrIbwzgkIWQTavO =rvVd -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org
participants (8)
-
Anton Aylward
-
Carlos E. R.
-
Carlos E. R.
-
Cristian Rodríguez
-
David C. Rankin
-
Dylan
-
Greg Freemyer
-
jdd