Ripro Server Troubleshooting
EtherShare
& OPI
Recommended Solution
- Start NFS on the station. Type
: "smit nfs", select "NFS",
select "Configure NFS on the
system",select "start NFS".
- Restart EtherShare.
Recommended Solution
Rebuild desktop of a corrupted File
System may cause a 3 digit error on the Server.
- Shutdown the system and
perform a reboot.
- Login immediately as root and
turn the EtherShare server off
"/usr/local/es/stop-atalk now".
- Do fsck (SciSmit)
on all volumes.
- Start the EtherShare server
(SciSmit).
If this does not work try to
isolate the problematic disk by:
- Login immediately as root and
turn the EtherShare server off
"/usr/local/es/stop-atalk now".
- Add a "#" before the
a line of one of the disks in the
afpvolumes file (in
/usr/local/es/conf).
- Start the EtherShare server
(SciSmit).
- Repeat the above step for all
disks until you know which disk you have to mark
and the EtherShare server can run normally.
- Backup the disk ,format it and
restore it.
To Prevent network problems on the
Macintosh or if your customer is already facing problems
such as not being able to save Photoshop files to the
Ripro Server after opening, make sure that all the
Macintoshes on the network are running the latest version
of the Network driver.
Recommended Solution:
On the Mac, select Control Panels
-> Network, Verify that AppleTalk Version is 58.1.5
and Ethertalk version is: 2.5.7.
If version is lower than that, ask
the customer to get the Network Software Installer
version 1.5.1 from Apple and install on all him Macs.
Recommended Solution
Do "empty trash" in Ripro
Admin (root user).
Recommended Solution
- Do rebuild desktop and empty
trash once a week.
- See the Network driver version
(Open Network from the control panel and watch
the message at the bottom of the window). If it
is lower than AppleTalk - 58.1.5 and
EtherTalk 2.5.7 use the upgrade supplied from
Apple (Network Software Installer) or Scitex.
- Verify that Ripros disks
are not full, using "df" Ripros
disks should not be filled to more than 97% of
disk capacity. Should a disk exceed this value a
message will be sent to the Ripros Terminal
and to the Ripro-messages file.
- In Photoshop when saving and
immediately canceling - file may disappear.
TO RECOVER A MISSING
FILE
- Try Rebuilding the desktop of
the Ripros disk.
- See whether the file exists on
the Ripro from a UNIX shell (use ls command).
- Try using Norton Disk Doctor
for "Unerasing" files.
Recommended Solution
- Search the Error Log for
messages from the atalkd module.
- Check for inconsistent
configuration of network numbers or zone names in
AppleTalk networks.
and no user visible in the
AppleShare, sharing option folder.
Recommended Solution
- User ID for user
"macuser" changed. Do "ls -l"
at UNIX prompt and note the User ID for files
created by user "macuser". Do
"smit users" and change the User ID to
"macuser".
- Check the "sharing"
parameters of the volume. The user and group
should be the same as the group accessing the
data on the volume.
Recommend Solution
Restart Mac with rebuild desktop
option "alt-apple".
Recommended Solution
Do "rebuild desktop" on
volume containing folder (Ripro Admin). The .rsrc
directory is probably missing
Recommended Solution
Check the owner and permission of
the directory and .rsrc subdirectory "ls -al".
Recommended Solution
- On the Ripro do "df"
to verify if the Filesystem is mounted. If
mounted, unmount from all the Macs and do Rebuild
Desktop using Ripro Admin.If not mounted go to 2.
- See if disk is available by
typing "lsdev -Cs scsi".If not
available check hardware and run
"cfgmgr". If available do
"fsck" and then "mount -a".
- The Mac OS and UNIX use
different ways to calculate file sizes. To view
the exact amount of disk space used from the
Macintosh, select the file in question in the
Finder and choose "Information" from
the File menu.
- The privileges for Unix
directories or files may have been changed from
within Unix.«
Even though the server is visible
in the Macintosh Chooser, volumes cannot be selected or
mounted and/or printing services cannot be selected.
Recommended Solution
- Check for error messages on
the Macintosh.
- Check for error messages from
the Unix System Error Log (from afpsrv/desksrv
and papsrv modules) regarding RPC, NFS, the file
system, and SCSI devices.
- Check if the "x"
(Execute) privilege may have been removed from
one of the UNIX directories (use "ls -l
/dataVolumes").
- Check that the number of
EtherShare licenses is not larger than the
permitted number.
Note: If a Macintosh drops
its connection to the file server or terminates it by
shutting down or rebooting, it takes about three minutes
before the EtherShare file server can reuse this license.
Recommended Solution
Check that file name is less than
27 characters
Recommended Solution
- Make sure you are not using
EtherShare from the Scitex LANC board on the Mac.
- Check total throughput of the
Server. Start "LanTest" on several Macs
on the network, select any of the Ripros
Volumes to test, repeat the test at least 10
times.
- Models 34H and 55L should give
total of about 800 KB/Sec, models 570 and 41T
about 1270 KB/Sec.
Note:
- Tests
should only be performed on local file
systems of the Ripro server. In no case
should imported NFS file systems be used for
tests, because there will be a performance
hit of 30-50% even under ideal condition.
- The
LanTest results may differ from the data
transfer rates observed in the
MacintoshFinder. LanTest achieves the maximum
data transfer rate possible between
Macintoshand server.
- Normal
Macintosh applications achieve slower
transfer rates because they have toprocess
the data as well as transferring them.
perform tests with FDDI or at least 2Ethernet
boards.
- Check versions of the
AppleTalk and EtherTalk drivers recommended
58.1.5 and 2.5.7. If drivers are not updated use
Apples "Network System Installer"
disk.
- If the first iteration of the
test shows good results that deteriorate in
subsequent iterations:
- The
server load may be too high
- The
server may be equipped with additional
RAM.
- An inconsistent configuration
of an AppleTalk network or physical problems in
the network (missing or improper termination,
cables that are too long or too short) may
adversely affect data transfer rates.
- Check if the problem occur
even when the Macintosh is booted without system
extensions that are not included with the
operating system.
- See if the Error Log contain
messages from the "atalkd" module?
- Consult your Network
specialist.
Recommended Solution
Please note that the new Macintosh
8500 / 9500 (or any other PCI Macintosh models) require
the "OpenTransport 1.0.7" software in order to
communicate with other devices.
When not installed, these Macs can
cause communication problems over an AppleTalk network
when accesing other devices (e.g. Ripro Server).
OpenTransport 1.0.7 is available
from Apple as an update to 1.0.6 which is included in the
Macintosh OS version 7.5.2GM.
If your Macintosh has an older OS
version than 7.5.2GM - Apple recommends to install it
from scratch.
Please contact your Apple dealer in
order to obtain this update.
Files or folders have the wrong
icons or appear in the wrong directories.
Recommended Solution
- No overlapping EtherShare
volumes must be defined.
- The UNIX privileges for the
".rsrc" directory may be insufficient.
- The desktop of the volume may
be defective and needs to be rebuilt using Ripro
Admin ("rebuild desktop").
Recommended Solution
- The desktop of the EtherShare
volume may be defective and needs to be rebuilt
using Ripro Admin.
- It may be possible to resolve
this problem by having the "root" user
mount the volume in question and deleting a file.
(It doesnt matter which file.) This creates
a new "Network Trash Folder" that is
also available to all other users.
Recommended Solution
- Check that file system is
mounted on the RS/6000 ("df") if not,
check that the disk is "available"
("lsdev -Cs SCSI").
- If the disk is
"defined", remove it ("rmdev -l
<hdiskx> -d") and do
"cfgmgr".«
When configuring a Ripro Server
with Multi Communication boards, Helios will publish the
Ripro on one of them only. It will always be the one that
has the driver name with the lowest letter.
For example:
Ripro Server with 1 ethernet board
(on board - 41T, or additional 5xx family) and 1 FDDI
bord. Setting the Helios (using netconf) to define cable
range and zones will create et0 and fi0 with the proper
parameters. Since e comes before f Helios will write the
et0 definition first and therefore will publish the Ripro
on et0 (something we normally do not want).
Recommended Solution
There are 2 ways to overcome this
problem:
- Do not define et0 for
ethershare (if only fi0 exist the ripro will be
published there).
- Correct the line order in the
configuration file. To do that:
- login as root.
- type: cd
/usr/local/es/conf
- type vi atalk.conf
- Change the atalkd
configuration from, for example:
atalkd:
if="et0:1-100:140:Zone1",
if="fi0:101-200:140:Zone2"
to:
atalkd: if="fi0:101-200:140:Zone2",
if="et0:1-100:140:Zone1
the all the communication
to/from the ripro will be on the fi0 from now
(Instead of the et0, before the change).
Note:
the next time you will run netconf again the file
will return to its original configuration and the change
via vi is needed again.
Reading or writing files or
displaying files or directories is not possible.
Recommended Solution
Make sure that the desksrv process
is running and check whether this process generates any
error messages. ("ps -e |grep desk").
Recommended Solution
- Unmount from all the Macs and
do Rebuild Desktop using Ripro Admin.
- Remove .Desktop and .Desksrv
files using "rm .Desk*" command. Use
"pwd" to verify correct position before
removing.
- Check permission of
/etc/hosts. Permission should be 664, use
"chmod" if not correct.
Recommended Solution
Check OPI search paths (Ripro
Admin).
Note:
Make sure you have the "OPI Setup" in the Ripro
Admin folder when opening Ripro Admin.
Recommended Solution
Call Apple for a new EPROM for the
printer.
When printing to a Ripro Server's
printer spooler fails due to lack of available disk
space, spool-space is needed to be increased.
Default size of /var (which is the
spooling file system) is about 100MB.
If an error has occured, additional
100MB chunk is recommended for the spooler.
Recommended Solution
- Login to the server as root.
- Type:
df
/var
The result
should look like:
| Filesystem |
Total KB |
free |
%used |
iused |
%iused |
Mounted on |
| /dev/hd9var |
102400 |
98188 |
4% |
102 |
0% |
/var |
Note the Total space
(KB equals 1024 blocs which is twice of 512 Blocks)
The above /var has a size of 204800 (in 512 blocks).
- To change (add 100MB or
200,000 units of 512K blocks), type:
chfs
-a size=+200000 /var
- To verify the change, type
"df/Var" again.
Note:
In case the new size is still not enough, an
additional chunk may be added. The system will notify in
case there is not enough free disk space for the
operation. This command is irreversible. Take into
consideration that the free space on the system disk is
limited and may be necessary for future use (fonts, etc.)
therefore add only what is a necessity.
|