Excessive use of memory.

[giuliox@xxxxxx (Giulio)]

Linux 2.0.33 64MB mem, 128 swap, leafnode 1.8.

For about a month leafnode worked ok.
Then it started sucking up resources in the final stage of a download (when rewriting
overview files).
Swap goes from 0 to 70MB and more.
Any idea?

for example:
...
corrected 76 lines in /var/spool/news/it/hobby/satellite-tv/digitale/.overview
wrote /var/spool/news/it/hobby/satellite-tv/digitale/.overview
....
and swap goes high

This is vmstat, every second; during the download of the messages swapd is 0 then as soon
as it reaches "wrote /var/spool..." it starts growing.
Then I killed fetch not to take the sistem down, but when I didn't know fetch was the
responsible the swap went over 70MB, and after killing fetch all processes remained
swapped out until we "used" them so the system was slow.

 procs                  memory    swap        io    system         cpu
 r b w  swpd  free  buff cache  si  so   bi   bo   in   cs  us  sy  id
0 1 0     0  1184  2864 16288   0   0  109    3  287  220  10   4  87
 0 1 0     0   612  2860 16564   0   0  485    0  385  471   3   5  92
 1 0 0     0   940  2860 16472   0   0  188  500  353  270   1   8  91
 1 0 0     0   764  2860 16632   0   0  316    0  358  341   3   8  89
 1 0 0     0   880  2852 16524   0   0  430    0  304  412   2  13  86
 1 0 0     0   780  2852 16600   0   0  207  500  256  296   4   4  92
 1 0 0     0   620  2852 16648   0   0  442    0  278  341   2   8  90
 1 1 2   316   604  2324  3720   0 316  107  537  290  422   4  25  71
 2 0 0   740   604  1660  3716   0 424    3  148  287  276   3   4  93
 2 1 0  1656   608  1660  3736   0 916   31  229  444  574   1   5  94
 1 1 1  2356   604  1660  3768   0 700   40  345  329  446   1   4  95
 1 1 1  3036   616  1392  3704   4 684   65  171  324  440   1   8  92
 1 1 0  3412   612  1392  3656   4 380  335   95  446  360   1  25  74
 2 0 0  4312   620  1392  3596   4 904   63  226  346  480   2  23  75
 2 0 0  5248   620  1392  3568  16 952   64  238  363  516   1   8  92
 2 0 0  5544   620  1388  3532   0 296   25  502  436  278   1   9  90
 2 1 0  6560   604   244  3588   0 1016   93  254  466  568   3   8  90
 2 1 0  7612   604   244  3564   0 1052   65  263  381  554   1  10  89
 2 0 0  8648   604   244  3532   0 1036   56  259  462  545   2  10  89
 2 0 0  9552   604   244  3564  20 924   83  231  397  568   2   8  90
 1 0 0 10256   604   216  3428   8 712  106  186  318  525   2   9  89
 1 1 0 11340   604   216  3376   4 1100   21  275  411  646   0  10  90
 2 0 0 12428   624   216  3356   0 1076   86  269  662  621   0  10  90
 1 2 0 13368   604   216  3248   0 940  230  235  394  636   2   8  90
 0 2 0 14308   604   216  3188   8 948  171  237  632  649   1  10  89
 1 3 0 15044   608   216  3200  36 788  110  206  527  496   3   7  90
..........

When I killed fetch the swap went down, but not to 0, and a lot of memory (leaving out
cache and buffers) was free (say 10MB intead of 600k); so it seems fetch used it.

For the above example I invoked fetch as
fetch -vv

The only weird thing in the log is texpire warning about non printable chars,
texpire[16320]: xover error: non-printable chars.
Feb  2 02:55:07 i3 texpire[16320]: bad overview line for it.comp.win95.software/
4011440

Solutions?

Thanks.

-- 
Giulio
giuliox@xxxxxx

-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list