|
Grex > Helpers > #138: Grex System Problems - Winter 2004/2005 |  |
|
| Author |
Message |
| 25 new of 870 responses total. |
keesan
|
|
response 258 of 870:
| 
Jan 6 15:55 UTC 2005 |
asd, aka 'smart' is using 97% of CPU for the past 550 minutes, running 'john'.
I will email gelinas. Grex is still usable. Is there some way to limit
individual users to 1 or at most 5% of CPU use?
|
keesan
|
|
response 259 of 870:
|
Jan 6 16:06 UTC 2005 |
I have in the past couple of days received three fragments of emails (presumed
spam) consisting of just the beginning of the header such as:
From the-concourse-on-high Thu Jan 6 11:01:04 2005
Received: from [201.17.23.27] (helo=c911171b.rjo.virtua.com.br)
by grex.cyberspace.org with smtp (Exim 4.42)
id 1CmZp2-0003qY-M4; Thu, 06 Jan 2005 10:45:13 -0500
Received: (from pyroxenite@201.17.23.27)
by helmholtz5[1
Is this just sloppy spam-writing or something going wrong in the middle of
the mail receiving process at grex? Or vandalism?
|
petercon
|
|
response 260 of 870:
|
Jan 6 16:07 UTC 2005 |
Some people may have more problems in their scripts now that we've
moved away from a SysV UNIX to a BSD UNIX - the "usr/ucb" directory in
a SunOS sytem is where BSD UNIX commands were put in Suns SysV OS.
Something like the move from Korn shell scripts to bash. Shell scripts
using Sun's SysV commands may not work the same in BSD (or be missing
entirely) so be aware.
Also, there are more differences in the directory structure and the
whole environment and deamon setup that may affect scripts written in a
SysV system. Better test your scripts before trusting them.
In my case (so far) I had to remove all my aliases and removed
the '/usr/ucb' reference in two places in my .profile. I also notice
git is gone and without mc or git Grex is not very easy to use as file
management is a pain without a file manager. I'll keep playing though.
|
tsty
|
|
response 261 of 870:
|
Jan 6 16:57 UTC 2005 |
general question: is there a way to spam-filter for plain-ol-mail?
pine is a pain (imnsho) even with its 'advanced' features (bloat?).
plain-0L-mail is soooooooooooooooooooo easy to use/learn adn it
soes not suck up f feast-full of quota blocks, either.
|
mcnally
|
|
response 262 of 870:
|
Jan 6 17:14 UTC 2005 |
re #261: the tool you want to use is probably procmail. you can
pass mail off to a filter before it is ever delivered into your inbox
and the read the mail that passes the filter with whatever MUA (mail
user agent, aka "mail reader") you want.
Sindi can probably tell you how to set up an elaborate system of
procmail rules to try to screen out spam or you can wait until staff
have time to install a system like SpamAssassin or other anti-spam
package and have procmail use that.
|
petercon
|
|
response 263 of 870:
|
Jan 6 17:20 UTC 2005 |
I don't see color in any command (ls, links, lynx, more(less), w3m,
etc.) anymore. I forced pine to use color. Still using Putty which
means TERM is set to xterm. On the Sun aliasing "ls -color" worked but
not on OpenBSD.
|
twenex
|
|
response 264 of 870:
|
Jan 6 17:38 UTC 2005 |
Set TERM to ansi. You can do this in PuTTY too.
|
keesan
|
|
response 265 of 870:
|
Jan 6 17:43 UTC 2005 |
A friend reports that Pine spellcheck is not working. I tried it and got a
message about something alternate and 255.
Please feel free to copy /a/k/e/keesan/.forward and also
/a/k/e/keesan/.procmailrc but change keesan in the latter file to your own
login and delete all lines starting with # and also change my whitelist (the
remaining parts with $MAIL on the last of three lines) to your whitelist by
putting in the addresses of friends who write you rather than the From:'s that
I have chosen to let through from my friends. And email keesan for help.
The complicated bit about Nigeria is to send Nigeria spams to a nigeria folder
and also to polygon who posts them at his website.
|
davel
|
|
response 266 of 870:
|
Jan 6 17:54 UTC 2005 |
Re 260: I think /usr/ucb is a Sun-ism. It's certainly not SYSV, and the SunOS
we were running before was a BSD, pre-Solaris SunOS.
|
drew
|
|
response 267 of 870:
|
Jan 6 22:01 UTC 2005 |
Re #253:
lsz works. One major issue solved, one (or two) more to go.
|
keesan
|
|
response 268 of 870:
|
Jan 6 23:26 UTC 2005 |
Party (hayz) is now splitting 98% of CPU usage with smart/asd. Could some
staff member kindly delete both accounts?
|
tod
|
|
response 269 of 870:
|
Jan 6 23:31 UTC 2005 |
load averages: 2.70, 2.73, 2.77
18:31:34
138 processes: 3 running, 135 idle
CPU states: 94.2% user, 0.0% nice, 5.6% system, 0.2% interrupt, 0.0% idle
Memory: Real: 71M/210M act/tot Free: 1301M Swap: 0K/3072M used/tot
PID USERNAME PRI NICE SIZE RES STATE WAIT TIME CPU COMMAND
20421 smart 64 0 5668K 5980K run - 953:53 49.37% john
11064 hayz3141 64 0 300K 896K run - 44:26 45.80% party
29978 _mysql 2 0 34M 17M sleep poll 1:29 0.00% mysqld
32358 _syslogd 2 0 164K 488K sleep poll 1:16 0.00% syslogd
3601 named 2 0 2516K 2852K sleep select 1:05 0.00% named
20957 exim 2 0 580K 696K sleep select 0:27 0.00% exim-4.42-2
15359 _pflogd 4 0 512K 328K sleep bpf 0:08 0.00% pflogd
15771 root 2 0 284K 1000K idle select 0:05 0.00% sshd
10834 root 2 0 1092K 1932K sleep select 0:04 0.00% httpd
|
mcnally
|
|
response 270 of 870:
|
Jan 7 00:49 UTC 2005 |
re #268: quoting % CPU usage is not enough to establish that a user
is abusing the system. At the very least we'd need to know the load
average on the system as well. When not very much is going on it's
not unusual for a single process or a few processes to appear to hog
the CPU. It doesn't necessarily mean they're starving other jobs.
|
keesan
|
|
response 271 of 870:
|
Jan 7 03:29 UTC 2005 |
Is it considered good manners to be logged on for 954 (minutes, hours?) using
this much CPU time? Smart appears to have been logged on since very early
Thursday (finger) so I don't really know what the TIME column means.
|
gull
|
|
response 272 of 870:
|
Jan 7 03:32 UTC 2005 |
That's 954 minutes of CPU time. It's the cumulative amount of time
that process has used the CPU, which is not necessarily the same as
the amount of time since it was started. (In fact, unless the process
has continuously used 100% of the CPU, it will always be lower.)
|
keesan
|
|
response 273 of 870:
|
Jan 7 04:38 UTC 2005 |
Is that not a lot of CPU time to be used by one user? Until recently that
process was using about 97% of cpu time. Could you take a look at what is
going on? User smart lists as name 'asd' which is one of many things on my
twit filter so this is likely to be some sort of 'joke'.
|
petercon
|
|
response 274 of 870:
|
Jan 7 15:22 UTC 2005 |
Yes /usr/ucb is a "Sun-ism". That stands for "University of California
at Berkeley", i.e. BSD's original "home". SunOS _is_ SysV, has been
since I started using it in 1988. You'll find SysV commands
in /usr/bin and BSD commands in /usr/ucb. You'l find the SysV
conventional directory setup including /dev/tty??? (which is not in
BSD). Here's a script (everytty - which lists all the ports in use by a
particular user) which I have run on the old Sun box - it's pure SysV
(will even work on Linux which follows SysV conventions). It won't run
under any BSD derivative. It's called by "everytty username".
#!/bin/sh
for a in `find /dev/tty[a-s]? -user $1 -print`
do
a=`basename $a`
echo "
$a:"
ps -xt$a
done
Sun did the /usr/ucb thing to add BSD-isms to its OS and allow users to
choose which "flavor" they wanted to see by changing the order of their
PATH, either putting /usr/bin or /usr/ucb first in their PATH. Hope
that clears things up.
|
twenex
|
|
response 275 of 870:
|
Jan 7 15:27 UTC 2005 |
http://kb.indiana.edu/data/agjq.html?cust=023362.83043.131
SunOS is a Sun Microsystems implementation of the Unix operating system.
Solaris is SunOS packaged with a number of additional tools, and a graphical
user interface (GUI) environment. Since Sun Microsystems did not offer the
Solaris product until SunOS 4, SunOS and Solaris have different version
conventions (e.g., Solaris 1 includes SunOS 4, while Solaris 2 includes SunOS
5). To further confuse the naming scheme, Sun now refers to Solaris by just
its point release (e.g., Solaris 7, 8, or 9 instead of 2.7, 2.8, or 2.9).
When Bill Joy, one of the main programmers of the Berkeley Software
Distribution (BSD), helped found Sun in 1982, he brought with him the elements
for the first release of SunOS. Up through version 4.1.x (Solaris 1.x), SunOS
remained a heavily BSD-influenced Unix implementation. However, in the late
'80s, Sun entered into a partnership with AT&T, which was then developing the
other major Unix flavor, System V. The result was System V release 4 (SVR4),
which incorporated BSD as well as SunOS extensions (e.g., NFS). Subsequently,
with its version 5.x (Solaris 2.x) releases, SunOS shifted from its BSD
origins to SVR4.
|
mfp
|
|
response 276 of 870:
|
Jan 7 15:46 UTC 2005 |
http://www.bsdforjesus.org/
|
rksjr
|
|
response 277 of 870:
|
Jan 7 17:00 UTC 2005 |
Our site:
Linkname: Grex Web Server Statistics
URL: http://www.cyberspace.org/stats/
is currently reporting for time ranges no later than:
October 03-October 09 2004
|
tod
|
|
response 278 of 870:
|
Jan 7 17:09 UTC 2005 |
UH OH
|
mfp
|
|
response 279 of 870:
|
Jan 7 17:14 UTC 2005 |
OH DEAR>
|
keesan
|
|
response 280 of 870:
|
Jan 7 17:24 UTC 2005 |
How do I turn on mail size limits in Pine? All I could find in Setup was
sorting by size. I just got a 176K 'file.zip' attachment purportedly from
China. In the old grex the per-mail limit was 100K and anything larger would
bounce then people would write to complain and be instructed to send me plain
text, or just one small jpeg not a 3MB one.
Thanks to whoever got the load averages back from 3 to 1. Speedy!
|
mcnally
|
|
response 281 of 870:
|
Jan 7 17:35 UTC 2005 |
> How do I turn on mail size limits in Pine? All I could find in Setup
> was sorting by size.
You don't. By the time Pine (or any other mail reader) gets to see
how large a message is it's too late; the message has already been
delivered to your mailbox and then read by your e-mail program.
If you want to reject all mail above a certain size you can easily
do that with procmail before it gets delivered to your inbox.
A recipe like "* > 100000" will match messages over 100,000 bytes.
Once you match them you can decide what you want done with them.
|
keesan
|
|
response 282 of 870:
|
Jan 7 17:59 UTC 2005 |
You mean something like:
* > 100000
/dev/null
(I am not sure if the :0: on my first line got into this response).
Why is grex no longer placing a limit on mail size? Someone else complained
recently about having to empty spams from his mailbox several times a day or
it would fill up with junk like this. Or maybe it is a virus.
I only know how to filter on headers (:0:) or message body (:0B:).
If grex is going to allow these large mails to get through now, could someone
possibly write up a script to simplify rejecting them with procmail, which
is not for beginners?
|