Etch is eating up CPU!
 

Hello,

A few days ago I installed Debian Sarge and upgraded it to Etch. Now, without any program open I have 97.9% of my memory (512MB) used and then some of the SWAP space, and as a consequence it is getting really slow.. And this is really odd because I have been using Fedora Core 3, Ubuntu, Mandrake and this has never happened before. The computer has an Athlon XP 2500 processor, 2 hard drives of 80GB each and 512MB RAM and a GeForce FX5200 with the nvidia drivers installed, so I think that it is not outdated to be used for Debian.

I don't know how to list the processes as they appear in KDE System Guard or Gnome System monitor, but there are plenty that are running under root, such as "getty" (there are 6 of them at this moment and can't kill them because they reappear), "kjournald" (7 of them) and X is using 160 MB of memory.

Any ideas?

In a teminal type

top

and see which processes are eating up your cycles.

The amount of memory X uses is essentially wrong because IIRC, it includes the video memory on your card as well as real memory being used by the X server.

Most likely memory isn't your problem, Linux always uses nearly all of your memory. This box here has 1G and there's 13M free. But 475M of that is cache which will be reclaimed if memory gets low so it doesn't matter.

Goodmorning!

Thank you all for your replies.

I still can't just copy/paste from terminal the output of top. But it's more or less: 1 user, load average: 0.46, 0.25, 0.15 Tasks: 82 total, 1 running, 79 sleeping, 1 stopped, 1 zombie
Cpu(s): 3.0%, 0.3% sy, 0.0 % ni, 94% id, 2.6% wa, 0.0% hi, 0.3% si
Mem: 516416k total, 511488 used, 4928k free, 107216k buffers
Swap: 1012960k total, 4036k used, 1008924k free, 165108k cached

and the processes that report more %MEM are
firefox (6.9%), amule (1.7%), XFree86 (1.7), kdeinit (2.9%)

None of the other distributions have I ever noticed that were using up all this memory, but this time I did notice it because the computer was getting somehow slow and there was a lot of disk activity every now and then.

Plus I forgot to mention that while top reports CPU load 3%, the graph on system monitor seems more like a cardiogram, going from 4% minimum to a maximum of42% when the computer is apparently idle.

There's only 4M of swap being used and over 160M of memory is cache, memory isn't your problem.

Ok, about the memory it makes sense to me, but what about the disk activity and the cpu load?

It's hard to say, run 'vmstat 1' while it's running slow and see what it says.

Copy/paste of one of the outputs:

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
1 0 4044 12256 9692 320452 0 0 0 0 1648 2164 16 3 81 0
0 0 4044 12256 9692 320452 0 0 0 0 1337 1793 29 2 69 0
0 0 4044 12256 9692 320452 0 0 0 0 1184 1561 26 4 70 0
0 0 4044 12256 9692 320452 0 0 0 0 1242 1478 12 3 85 0
0 0 4044 12256 9700 320452 0 0 0 96 1234 1602 17 4 79 0
0 0 4044 12256 9700 320452 0 0 0 0 1404 1882 24 2 74 0
0 0 4044 12256 9700 320452 0 0 0 0 1621 1984 24 3 73 0
0 0 4044 12256 9700 320452 0 0 0 0 1133 1131 5 0 95 0
0 0 4044 12304 9700 320452 0 0 0 0 1659 2061 11 2 87 0
0 0 4044 12304 9700 320452 0 0 0 4 1718 2002 10 4 86 0
0 0 4044 12176 9704 320472 0 0 24 0 1538 1851 38 11 50 1
0 0 4044 12240 9728 320472 0 0 0 128 1597 1856 13 1 86 0
0 0 4044 12200 9728 320472 0 0 0 0 1529 1963 33 4 63 0
0 0 4044 12200 9728 320472 0 0 0 0 1634 2066 17 1 82 0
0 0 4044 12152 9728 320472 0 0 0 0 1520 2704 68 10 22 0
2 0 4044 12152 9728 320476 0 0 0 0 1696 3904 59 10 31 0
3 0 4044 12136 9736 320476 0 0 0 76 1527 3619 59 10 31 0
2 0 4044 12136 9736 320476 0 0 0 0 1655 3226 62 16 22 0
0 0 4044 12136 9736 320476 0 0 0 0 1476 1774 13 1 86 0
0 0 4044 12136 9736 320476 0 0 0 0 1587 1714 9 2 89 0
0 0 4044 12136 9736 320480 0 0 0 0 1517 1857 24 2 74 0

But I don't know how to interpret this

Please repaste that inside of code blocks so that the formatting stays readable.

procs -----------memory--------------- ---swap-- -----io---- --system--- ----cpu----
r b swpd free buff cache si so bi bo in cs us sy id wa
1 0 4044 10388 11296 312872 0 0 0 29 1197 1806 36 6 58 0
2 0 4044 6996 11296 312872 0 0 0 0 1247 1864 32 5 63 0
1 0 4044 10192 11296 312880 0 0 0 0 1578 2363 60 13 27 0
0 0 4044 10132 11304 312904 0 0 0 104 1599 2440 41 9 50 0
3 0 4044 9236 11304 312912 0 0 0 0 1405 2299 42 8 50 0
2 0 4044 10068 11304 312920 0 0 0 104 1858 2399 65 7 28 0
0 0 4044 10068 11304 312920 0 0 0 0 1267 1541 20 2 78 0
2 0 4044 5428 11304 313180 0 0 240 0 1560 2310 43 8 45 5
0 0 4044 9540 11304 313196 0 0 0 0 1594 1828 37 3 60 0
0 0 4044 9544 11312 313196 0 0 0 132 1143 1112 10 2 88 0
1 0 4044 9544 11312 313196 0 0 0 6 1196 1170 9 3 88 0
0 0 4044 9544 11312 313196 0 0 0 0 1145 1086 8 1 91 0
0 0 4044 9544 11320 313196 0 0 0 16 1095 954 8 4 88 0
0 0 4044 9544 11320 313196 0 0 0 3 1110 950 8 1 91 0
1 0 4044 9544 11320 313196 0 0 0 0 1153 980 10 1 89 0
0 0 4044 9584 11324 313196 0 0 0 5 1103 922 10 4 86 0
0 0 4044 9584 11324 313196 0 0 0 0 1154 1031 8 1 91 0
1 0 4044 8688 11324 313196 0 0 0 0 1095 882 8 2 90 0
0 0 4044 9584 11324 313196 0 0 0 0 1105 946 9 2 89 0
0 0 4044 9616 11324 313196 0 0 0 0 1122 936 10 2 88 0
0 0 4044 9616 11324 313200 0 0 0 4 1732 1601 15 2 83 0

Feww! Did it!

Now what!!! I spent half an hour reformating the thing by hand and it became a mess again!

Do you have DMA enable on your hard disks? To check run hdparm

"hdparm /dev/hdxx" Fill in the xx to what it is for you. If that's not the case you can set DMA on with hdparm like this:

"hdparm -d1 /dev/hdxx"

Yes, dma is on, no problems there.

Well then you either didn't run the vmstat you posted when it was running slow or something is weird, because that vmstat output show's no swapping at all and very little disk I/O.

Are you sure it couldn't just be X that's slow? i.e. window movements, redraws, etc.