Smart questions
Smart answers
Smart people
INTELLIGENT WORK FORUMS
FOR COMPUTER PROFESSIONALS

Member Login




Remember Me
Forgot Password?
Join Us!

Come Join Us!

Are you a
Computer / IT professional?
Join Tek-Tips now!
  • Talk With Other Members
  • Be Notified Of Responses
    To Your Posts
  • Keyword Search
  • One-Click Access To Your
    Favorite Forums
  • Automated Signatures
    On Your Posts
  • Best Of All, It's Free!

Join Tek-Tips
*Tek-Tips's functionality depends on members receiving e-mail. By joining you are opting in to receive e-mail.

Posting Guidelines

Promoting, selling, recruiting, coursework and thesis posting is forbidden.
Jobs from Indeed

Link To This Forum!

Partner Button
Add Stickiness To Your Site By Linking To This Professionally Managed Technical Forum.
Just copy and paste the
code below into your site.

FLAdmin (MIS) (OP)
6 Mar 06 16:17
First of all I'm a windows admin with just enough knowledge to get me around Unix a bit for small things. However we run our main application wich is FoxPro for Unix based.

The problem we have is Intermittent Hangs to the point where it takes about 1-2 minutes to see any kind of screen update. My only solution has been rebooting the machine unfortunately. I wish I could give more information but if anyone can point me in the right direction I would be happy.

I would like to know also how to do a chkdsk equivalent type command to check for bad sectors on the disk without harming any data.

Also does this thing have any simple path or order to start troubleshooting, maybe a log? and is there any gotchas to watch out for that are worth knowing. thanks

This Machine is: (incredibly)
3.6 Gz P4
1GB RAM
HD, unsure. I believe 2, one for backups.
motoslide (MIS)
6 Mar 06 17:57
Logs:  /usr/adm/messages and /usr/adm/syslog
These can get long, so you may try looking at each with the "tail" command right after you notice a problem.

"fsck" is similar to chkdsk.
# man fsck  (for instructions)

If you had any bad sectors, there would be entries in /usr/adm/syslog.  

That sounds like new hardware.  Which version of OpenServer are you running?
# uname -X  

When the system is "hung", is the console also unresponsive?  

You can run a command like this to help determine which process is hogging the CPU when the performance is poor:

CODE

ps -e -o "pcpu" -o "pid=" -o "time" -o "user=" -o "args=" |
awk '$1 > 5' |
sort -r -n

edfair (TechnicalUser)
6 Mar 06 19:33
badtrack or scsibadblk do some scans

to see what you have for hard drives:  l /dev/hd*  (l/c L)

to see how the drives are set up

divvy /dev/hd0x  or  /dev/hd1x   (use 1 point from previous)

to see what you have for filesystems

df -k -v

for info on any of these  man command

Ed Fair
Give the wrong symptoms, get the wrong solutions.

stanhubble (MIS)
7 Mar 06 7:59
you should also find out what version of OS you are running.

uname -v     
uname -X will give you more information.

if it is not at least 5.0.6a then really need to update it,
P4's and anything prior to 5.0.6a don't get along very well.

sar will tell you where the bottlenecks could be.
cdlvj (MIS)
7 Mar 06 8:12
>>>if it is not at least 5.0.6a then really need to update it, P4's and anything prior to 5.0.6a don't get along very well.

Those versions do not support hyperthreading, and probably is the root of your problem
FLAdmin (MIS) (OP)
7 Mar 06 10:52
Thanks guys, great help. so far this is what I got. maybe you'll sniff something funny out of it.

System = SCO_SV
Node = sco
Release = 3.2v5.0.6
KernelID = 2000-07-27
Machine = Pent4
BusType = ISA
Serial = 2DJ031060
Users = 105-user
OEM# = 0
Origin# = 1
NumCPU = 1

brw-------   1 sysinfo  sysinfo    1,  0 Sep  5  2004 /dev/hd00
brw-------   1 sysinfo  sysinfo    1, 15 Sep  5  2004 /dev/hd01
brw-------   1 sysinfo  sysinfo    1, 23 Sep  5  2004 /dev/hd02
brw-------   1 sysinfo  sysinfo    1, 31 Sep  5  2004 /dev/hd03
brw-------   1 sysinfo  sysinfo    1, 39 Sep  5  2004 /dev/hd04
brw-------   1 sysinfo  sysinfo    1, 47 Sep  5  2004 /dev/hd0a
brw-r-----   1 dos      sysinfo    1, 48 Sep  5  2004 /dev/hd0d
brw-------   1 sysinfo  sysinfo    1, 64 Sep  5  2004 /dev/hd10
brw-------   1 sysinfo  sysinfo    1, 79 Sep  5  2004 /dev/hd11
brw-------   1 sysinfo  sysinfo    1, 87 Sep  5  2004 /dev/hd12
brw-------   1 sysinfo  sysinfo    1, 95 Sep  5  2004 /dev/hd13
brw-------   1 sysinfo  sysinfo    1,103 Sep  5  2004 /dev/hd14
brw-------   1 sysinfo  sysinfo    1,111 Sep  5  2004 /dev/hd1a
brw-r-----   1 dos      sysinfo    1,112 Sep  5  2004 /dev/hd1d

LIVE # /ivvy /dev/hd01
+-------------------+------------+--------+---+-------------+------------+
| Name              | Type       | New FS | # | First Block | Last Block |
+-------------------+------------+--------+---+-------------+------------+
|                   | EAFS       |  no    | 0 |            0|       20479|
|                   | NON FS     |  no    | 1 |        20480|      806911|
|                   | HTFS       |  no    | 2 |       806912|     8938495|
| (label = tmpinv)  | DTFS       |  no    | 3 |      8938496|    14058495|
|                   | DTFS       |  no    | 4 |     14058496|    30441461|
|                   | HTFS       |  no    | 5 |     30441462|    35561461|
|                   | NON FS     |  no    | 6 |     35561462|    35561471|
| hd01              | WHOLE DISK |  no    | 7 |            0|    35564527|
+-------------------+------------+--------+---+-------------+------------+
35561472 1K blocks for divisions, 3056 1K blocks reserved for the system


LIVE # divvy /dev/hd00
+-------------------+------------+--------+---+-------------+------------+
| Name              | Type       | New FS | # | First Block | Last Block |
+-------------------+------------+--------+---+-------------+------------+
| boot              | EAFS       |  no    | 0 |            0|       20479|
| swap              | NON FS     |  no    | 1 |        20480|      806911|
| root              | HTFS       |  no    | 2 |       806912|     8938495|
| tmpinv            | DTFS       |  no    | 3 |      8938496|    14058495|
| user2             | DTFS       |  no    | 4 |     14058496|    30441461|
| tmp               | HTFS       |  no    | 5 |     30441462|    35561461|
| recover           | NON FS     |  no    | 6 |     35561462|    35561471|
| hd0a              | WHOLE DISK |  no    | 7 |            0|    35564527|
+-------------------+------------+--------+---+-------------+------------+
35561472 1K blocks for divisions, 3056 1K blocks reserved for the system



 50.09   457    00:00:03     root /usr/unform60/rt/pvx /usr/unform60/uflist.pv -
arg start
 18.59  2959    00:00:00   bernie /usr/lib/foxpro/foxr.pr -t -i -o -e  selcompx
  7.55   206    00:00:02     root htepi_daemon /tmp
FLAdmin (MIS) (OP)
7 Mar 06 10:54
Mount Dir  Filesystem              blocks      used      free   %used
/          /dev/root              8131584   3221991   4909593    40%
/stand     /dev/boot                20480     11950      8530    59%
/tmpinv    /dev/tmpinv            5120000   3931808   1188192    77%
/user2     /dev/user2            16382966  13307310   3075656    82%
/tmp       /dev/tmp               5120000    179608   4940392     4%
/usr2      /dev/usr2             32489472  26843477   5645995    83%
/usr3      /dev/usr3              3072000   1134574   1937426    37%
/josedsk   lx:/home/public/PERF  47968136  37042640  10925496    78%
/usr/archi lx:/home/public/arch  47968136  37042640  10925496    78%
motoslide (MIS)
7 Mar 06 11:18
Your Unform process shouldn't be hogging all the CPU like that.  It's a quick form-generating utility.  This will, of course, consume CPU time while it is processing, but that should be short-term.  See if it continues to be at the top of that list.  You can enable debugging in Unform 6 if it proves to be the issue.  My guess is that you just happened to run that PS command while it was active.

Check that you have the recommended patches for that 5.0.6 O/S.  As mentioned by others, there are compatibility issues with the P4.  To the extent, (I've heard, but never seen) that it can cause the P4 to fail.
To see the O/S patches, run "custom".
motoslide (MIS)
7 Mar 06 11:21
Also, those last two filesystem entries are NFS-mounted filesystems (right?).  See if you can determine their purpose, and also see if there are files being modified or copied to/from those areas while the system is sluggish.
FLAdmin (MIS) (OP)
7 Mar 06 11:32
Here's whats in Custom:

3Com EtherLink PCI (ver 5.0.6b)
Bash - GNU Bourne-Again SHell (ver 2.03)
Broadcom BCM5700 NetXtreme Gigabit Ethernet (ver 5.0.7d)
Bzip2 - block-sorting file compressor/uncompressor (ver 0.9.5d)
DB - the Berkeley Database Library (ver 1.85.4)
Expect - programmed dialogue (ver 5.30)
FSU Pthreads (ver 3.5)
GCC - GNU Compiler Collection (ver 2.95.2pl1)
GNU Ghostscript, a PostScript and PDF interpreter (ver 6.51)
GNU Libtool, a generic library support script (ver 1.3.5)
GNU Privacy Guard (ver 1.2.4)
Groff - GNU nroff, troff and related tools (ver 1.15)
LDAP - Lightweight Directory Access Protocol (ver 3.3)
Lxrun - Linux Application Environment Emulator (ver 0.9.4pre3)
OSS631B - Supplemental Graphics, Web and X11 Libraries (ver 1.2.1)
OpenSSH - secure shell client/server (ver 3.1p1)
PRNGD - Pseudo Random Number Generator Daemon (ver 0.9.23)
Pacific CodeWorks, Inc. Olympus TuneUp (ver 6.0.0P)
Qpopper - POP3 protocol server (ver 3.1)
Redhat Package Manager (ver 4.0.4)
SCO OpenServer Enterprise System (ver 5.0.6j)
SCO SendMail (ver 8.11.0)
SCO Skunkware 2000 (ver 2000.1)
zlib - unencumbered lossless data-compression library (ver 1.1.4)
OSS638A: rarpd Supplement for OpenServer 5 (ver 1.0.0)
OSS640A: BIND Update (ver 1.0.0)
OSS642a - Cron supplement (ver 1.0.0)
OSS643A: Socket Driver Supplement for OpenServer 5 (ver 1.0.0)
OSS644B - Reboot supplement for OpenServer 5 (ver 1.0.0)
OSS645A: Audit Subsystem Security Supplement (ver 1.0.0)
OSS646B - Execution Environment Supplement (ver 1.1.0j)
OSS646C - Execution Environment Supplement (ver 1.2.0a)
OSS648A: Processor Supplement for OpenServer 5.0.6 (ver 1.0.0)
OSS650a - lpd supplement (ver 1.0.0)
OSS651A: Intel Microcode Update for OpenServer 5.0.6 (ver 1.0.0)
OSS660A - getty Supplement for OpenServer 5.0.6 (ver 1.0.0)
OSS663A - LPD Supplement for OSS646 (ver 1.0.0)
RS506A: Release Supplement for SCO OpenServer Release 5.0.6 (ver rs
RS506A: Software Manager Supplement (ver rs506a)
FLAdmin (MIS) (OP)
7 Mar 06 11:35
As for the NFS Drives, If I'm not mistaken... I think they are the windows friendly shares we have. because we do share the entire set of tables for the foxpro database in order to run a windows based fox pro app that we are developing as well. However I have been trying to import data with SQL, I considered this. but even after shutting down sql it was sluggish. for now mysteriously its fast again after a hard boot that made a disk checking come up at boot.
FLAdmin (MIS) (OP)
7 Mar 06 11:43
The system is slow now:

 54.84     2    00:00:06     root vhand
 12.41  2712    00:00:20    erika /usr/lib/foxpro/foxpro.pr -t -i -o -e  selcomp
x
  8.05  6466    00:00:00    erika mv 1115_HELLECOMFO_F-92834_060307_AHARON_SHA /
tmpinv/pdfinv-hist
FLAdmin (MIS) (OP)
7 Mar 06 11:44
Also at the Console:

Table_Grow excecdata table page limit of 50 pages(MAXEXECARGS) Exceeded by 1 pages
FLAdmin (MIS) (OP)
7 Mar 06 12:42
heres another snapshot:
 25.80  9604    00:00:34     jose popper
 22.28     2    00:00:21     root vhand
 15.08  9710    00:00:01     root /usr/lib/foxpro/foxpro.pr -t -i -o -e  selcomp
x
  6.90   199    00:00:05     root dtdaemon
  5.38 18974    00:00:55  vanessa /usr/lib/foxpro/foxpro.pr -t -i -o -e  selcomp
x
motoslide (MIS)
7 Mar 06 12:59
Make sure there exists a mailbox for any PC's which are configured to check for mail using the POP3 daemon (popper).  If you do not have a corresponding file in /usr/spool/mail, then popper gets very cranky.  I think it is generating the console messages.

You can create those files using the "touch" and "chown" commands:
# cd/usr/spool/mail
# touch user1
# chown user1 user1
# chmod 600 user1
(repeat as needed for the other users)
FLAdmin (MIS) (OP)
7 Mar 06 13:24
there is only less than 10 office people that use it to check mail, wich are there. there is also about a hundred or so for other users wich are not being used.
motoslide (MIS)
7 Mar 06 13:37
Are any of the mailbox files extremely large?
When a PC checks for Email, it will launch the popper daemon, which will start by creating a copy of the user's mailbox file, then transfer tne contents of the copy to the User's Email program.  Once that has been accomplished, it will copy the file back to the original name.  If the file is large (Megabytes), you could have a delay.  You've got new hardware, so disk I/O should be extremely fast.  

Before getting too worked up, you will need to see if there is a direct correlation between the processes which are active while the system is noticeably slow.
FLAdmin (MIS) (OP)
7 Mar 06 13:54
Here's the top bunch. they dont seem all that large.. 3mb? if thats correct

Size
2838531    Mar    7    13:23    root
2180044    Nov    24    13:23    marcelo
2131948    Dec    16    19:10    michelle
1389424    172    Mar    7 13    jose
1161297    Nov    14    2003    kristina
862117    Nov    28    11:19    terre
834011    May    2    2003    xfer
819592    Sep    27    2004    jeff
784361    Apr    17    2003    jessica
765960    Dec    21    11:21    alex
746549    Apr    17    2003    irene
723066    Apr    28    2003    info
718852    Jan    26    12:34    mary
718771    Jan    27    2004    linda
718567    Apr    25    2003    larry
FLAdmin (MIS) (OP)
7 Mar 06 14:01
You guys mentioned updating to 605a. Does Unix have some kind of auto update from ftp tool or anything?

and more importantly can they impact the FoxPro for Unix application?
stanhubble (MIS)
7 Mar 06 14:01
also see if it is cpu io or memory that the system is busy with.

sar 5 5

sar -r 5 5

FLAdmin (MIS) (OP)
7 Mar 06 14:04
SCO_SV sco 3.2v5.0.6 Pent4    03/07/2006

14:03:30   freemem   freeswp availrmem availsmem (-r)
14:03:35    120060   1520472    128631    308242
14:03:40    120064   1520472    128631    308242
14:03:45    120061   1520472    128631    308238
14:03:50    120045   1520472    128627    308182
14:03:55    120043   1520472    128627    308182

Average     120054   1520472    128629    308217

FYI I believe the Box has 1 GB RAM but i will doublecheck next reboot.
PHV (MIS)
7 Mar 06 14:35
The system is already patched as custom said to you:
RS506A: Release Supplement for SCO OpenServer Release 5.0.6 (ver rs
RS506A: Software Manager Supplement (ver rs506a)

Hope This Helps, PH.
Want to get great answers to your Tek-Tips questions? Have a look at FAQ219-2884 or FAQ181-2886

motoslide (MIS)
7 Mar 06 14:50
Since VHAND has been listed twice, you may benefit by decreasing some kernel parameters which deal with Disk buffers.
What is the output of these commands:
# sysdef|grep BDFLUSH
# sysdef|grep NAUTOUP

These are parameters which tell the O/S how often to check the disk buffer for pending write operations, and how old a "dirty buffer" should be before it is written back to disk.  Your Foxpro application vendor may have recommendations for these.  In the older days where CPU and DISK were slow, we achieved better overall performance by letting the buffers hang around for awhile before writing them out to disk.  With current hardware, it is often better to check these and flush them to disk often, so each pass only has to write a relatively small amount of information.  

Please read and review the information in this link:
http://www.aplawrence.com/SCOFAQ/scotec9.html

It sounds like it may reflect your symptoms.  Some of the recommendations might be beyond your comfort level, so proceed with caution.
motoslide (MIS)
7 Mar 06 15:01
I must be getting old and tired.   On that link I provided above, the section which might apply is the last regarding  "vhand spinning".  
stanhubble (MIS)
7 Mar 06 15:33
ok its not memory....but what about the first one

sar 5 5
FLAdmin (MIS) (OP)
7 Mar 06 17:08
I have Olympus tuneup and It looks like POPPER is using up A GIG OF RAM and RISING...!! and the user "Jose" as well as no one else can check their pop mail...  how to i tame this beast..??? the system is most likely crawling because of it
FLAdmin (MIS) (OP)
7 Mar 06 17:31
It looks like it killed itself. after reaching about 1.2GB of ram usage. Now his outlook keeps asking for POP user pass even though it is correct, yet it works perfectly fine for any other user. He is sucsessfully logged into a telnet session to sco with the same user pass right now.

do you know if popper maybe only likes to be checked for mail in certain time intervals or anything?
motoslide (MIS)
7 Mar 06 18:30
Kill his OUTLOOK (or whatever process on the PC is checking the mail)
Go to /usr/spool/mail
Look for all instances of "jose":

CODE

# ls -al | grep jose
-rw-------   1 jose     mail         485 Mar  7 15:24 .jose.pop
-rw-------   1 jose     sys            0 Mar  7 15:24 jose
(In the example above, this is a very small mailbox)
You are probably bouncing his Email between /usr/spool/mail/jose and /usr/spool/mail/.jose.pop

Once that has settled down, it will just sit in /usr/spool/mail/jose.  Check the size.  If it's big, you probably have no "EASY" choice but to clear it out.
Other options are quite tedious, or require you to configure some other mail reader program.  
If you wish to clear it out, just do this (there is no undelete option here, so make sure you are OK with losing all his pending mail):
# >jose
FLAdmin (MIS) (OP)
7 Mar 06 18:56
here's the deal... He's the boss. so i gotta be a little careful. However most emails are automated over time to his inbox so he gets well.... hundreds some days. I've seen it sometimes pull 26mb - 50mb into his outlook in a single gulp. luckly i moved all the user emails to an exchange server. I have not however been able to correctly redirect some email.  eg.  my exchange server relay's email for user@flmiami.com but not user@scosysv@flmiami.com my solution for now is to have 2 accounts in their outlook. one for exchange and the other for pop3 from unix for internally generated emails. and the output goes directly to exchange. nifty when it works.

i gotta think hard about what to do on this one... unfortunately more emails pile up every second.
motoslide (MIS)
7 Mar 06 19:36
I would recommend you try a product on the PC which will display the Email list without actually pulling it in.  This will allow you to selectively delete the ones which may be huge (and hopefully unwanted), yet preserve the ones of importance.  I use Mailwasher (www.firetrust.com).  I'm sure there are others.  You would still have the problem the first time it connects to the server to check mail, as it launches the popper daemon just like Outlook, and his Email will get copied.  But, once that goes through, you may be able to mark the ones you want to delete.  Outlook will timeout before it ever completes, which just puts you in this on-going cycle.
If you DO run such a utility, change his OUTLOOK settings to NOT automatically check for incoming mail every x-minutes.  Make that a manual process so you don't have 2 client programs competing for the same source mailbox.

You can also use the built-in "mail" command from a UNIX shell (while logged in as "jose").  That would probably be a better option, but the interface is awful.  Try if first as a user who doesn't have such a large inbox.  

Another option is to install PINE on the SCO box.  This has a better user interface that the built-in "mail" command.  

Just curious:  How big is his pending mail file?

FLAdmin (MIS) (OP)
7 Mar 06 19:36
okay I trashed his mail file with the ">jose" but he still cant check his pop account. it says its being used by another user... ?
FLAdmin (MIS) (OP)
7 Mar 06 19:38
By the way his file grew to 1.4 GB in about a  day or two. its insane. I ftp'd it out to a windows box and zeroe'd it.
motoslide (MIS)
7 Mar 06 20:02
The POP account problem will eventually go away.  Once again, it's probably in the "copy jose to .jose.pop" or "copy .jose.pop to jose" process.  This annoyingly occurs every time his PC tries to look for new mail.  Once Outlook gives up, the popper daemon has to copy .jose.pop back to jose.  Leave his Outlook OFF and make sure there is no "popper" process running for his user account.  Then re-check /usr/spool/mail/jose for size.  You can also make sure there are no processes running on that file:
# fuser /usr/spool/mail/jose

Make sure he is the owner of that file.  Also check for a /usr/spool/mail/jose.lock (might be jose.lck).  This is created as the mail transport tries to add to his mailbox.

You may have to develop a script which will monitor his mailbox and alert you when it gets above a certain size.  I'm not an Exchange expert, but I don't think it would be really happy with such large Emails either.

You may still be able to reduce the impact all of this has on the Foxbase users by reducing those Kernal parameters I had mentioned earlier (NAUTOUP and BDFLUSH).  If you set each to "1", then the system will chug along updating the disk every second, but with much shorter writes.  
FLAdmin (MIS) (OP)
7 Mar 06 20:25
Is it feasable to just ">jose" the /usr/spool/mail/jose file as well as the .jose.pop file? I'm trying to delete all this. The system has been hanging alot today.

Most notibly when I Opened up startx and checked any email regarding the lp. it just locks upon touching those, so i deleted all the emails there. also as soon as I opened the display adapter choose dialog. it immediately froze
FLAdmin (MIS) (OP)
7 Mar 06 20:53
well deleting that .pop file did the trick for that. He can check his email now.

however I cant explain why it was over a gigabyte of inbox so quickly since SMTP traffic does not come from the outside to that box...
motoslide (MIS)
7 Mar 06 21:51
With his permission, grab a copy of his mailbox file when it is of reasonable size and look at the mail headers.  You should be able to quickly see where the mail originates.  

I think you've identified your system performance problem.
You've passed the Unix SysAdmin initiation test.
FLAdmin (MIS) (OP)
7 Mar 06 21:57
Thanks man. It's like a cryptic dos. luckily dos 2.0 was where i started. Today I pretty much figured out how mail works, monitoring processes, identifying hogs, Qpopper is evil and I must try not to use it if possible. the hosts file missing server entry delayed emails. thanks guys.

I wrote a windows app to quickly view log files from a unix box with a simple drop down menu to choose different logs and an excel like interface with a single button click. I'll post the link when I unhard code the user/pass and let it use an INI instead so you can add locations of your own logs... its really handy.

Reply To This Thread

Posting in the Tek-Tips forums is a member-only feature.

Click Here to join Tek-Tips and talk with other members!

Back To Forum

Close Box

Join Tek-Tips® Today!

Join your peers on the Internet's largest technical computer professional community.
It's easy to join and it's free.

Here's Why Members Love Tek-Tips Forums:

Register now while it's still free!

Already a member? Close this window and log in.

Join Us             Close