3/7/2005 Changes in CSC 4-1-24
- Contact 27578078 reports truncation of the USERID configuration parameter.
This is corrected in this release.
- A CSC hang occurred during internal testing and another was reported by
contact 27170061. The root causes of these were never discovered but they
were linked to failures in CDI and CSCUI respectively. The code in CSC 4-1-24
is rearranged to make it less susceptible to failures in external programs.
- During internal testing with CSC 5R1 and 4R1, a condition was observed where
CSC 4R1 could not mount or dismount tapes and produced no error indications.
This occurs if CSC 5R1 using CARTTAPELIB$ is taken down and replaced with CSC
4R1. CSC 4-1-24 now produces a message during initialization for each drive
that it cannot use because that drive is still under CARTTAPELIB$ control.
11/9/2004 Changes in CSC 4-1-23
- UCF 25357435 reports a problem when the fields of the SIGNON configuration
parameter are too long. The improper truncation of long names is corrected
in this release.
- During internal testing a duplicate dismount message occurred and was traced
to incorrect handling of duplicate server responses. This is corrected.
- During internal testing a path unavailable message often occurred immediately
before the path is activated. Status handling is updated to eliminate this
false failure message.
- During internal testing it was found that setting the AUTOMOUNT parameter to
false disables BEFORE-MOUNT notifications. This is corrected.
- The sample JCL in NCS-PROCS is updated to prevent an SLS1203I error that can
occur after reconfiguration of the NCS server.
7/1/2004 Changes in CSC 4-1-22
- UCF 23277373 reports TIFSYNCH repeatedly tries to scratch a mounted VSM
volume and fails. Code is removed from CSC that was a workaround to a
problem that existed in CSC 3R4.
- The supporting material for UCF 23343975 showed that CSC does not complete
initialization if CSC selects one path through an interface and the server
returns its responses via another path through the same interface. CSC now
ensures that when a path is selected for use, all paths through that
interface are initialized and ready for use.
- Internal logging code in CSC is enhanced to ease the analysis problems
similar to those corrected in this release.
4/7/2004 Changes in CSC 4-1-21
- During internal tests, it was found that CSC would not initialize and use a
TSAM interface if CMS or CPCOMM was started after CSC. The interface would
also become unusable if CMS or CPCOMM was restarted while CSC is active.
These problems are corrected.
3/11/2004 Changes in CSC 4-1-20
- UCF 98906510 reports a CSC hang that occurred after a communication problem
that allowed session establishment but prevented data transmission. CSC now
times the data transfer portion of communication sessions and aborts
connections that time out.
- A coding error in CDI is corrected that caused a real time hang.
- Timing and internal buffer handling were adjusted to circumvent a CSC abort
that occurred after an extended server outage.
- The port used for CSC responses will be adjusted at CSC run time. This
allows CSC to function even if another program is using port 7000.
- The program CSCINFO/O is added to the CSC-2 file to display CSC product
registration information that will be needed to obtain a CSC 5R1 license key.
- A coding error is corrected that occasionally produced an incorrect
"parameter too long" error while processing the CSC configuration data.
12/19/2003 Changes in CSC 4-1-19
- Contact 98308933 reports a CSC failed to switch paths when Library Station
was taken down on one server platform and started on another. The path check
logic in CSC is updated to try alternate paths even when the server appears
to be reachable on the active path.
- UCF 9848430 reports that CDI cannot communicate with the high availability
server following a fail over. CDI is changed to recognize and processes the
network message telling that a new mapping of an IP address to a MAC address
has been established.
- UCF 98850091 reports an assignment failure when the requested volume resides
in an ACS with no drives that are up. CSC now will not direct assignments to
an ACS with no drives that are up. OS 2200 will then select a unit in the
run default ACS. The assignment will still fail, however, if there are no
drives up in the run default ACS.
- UCF 99014753 reports a CSC guard mode caused by the server returning an
undefined state code in response to a *CSC QUERY ACS keyin. A defect in the
CSC code that displays undefined server states is corrected.
- UCF 9903839 reports multiple restarts of CSC following a reload of the UCS
RTS common banks. A code defect in CSC is corrected that allowed each
activity that received the common bank reload contingency to schedule another
restart of CSC.
8/21/2003 Changes in CSC 4-1-18
- UCF 19723623 reports periodic CDI aborts on multiple systems. These occur
when CDI cannot create a temporary routing entry to respond to a PING or
redirect request. CDI is updated to shorten the life of temporary routing
entries and to discard responses rather than aborting if a temporary routing
entry cannot be created.
- UCF 95915060 reports an additional line in the volume report output that did
not exist in CSC 3R4. This line is an echo of the server command requesting
the volume report. It is now eliminated from the volume report output.
- Contact 57414735 reports a truncation error when the ACOBCSCUI interface is
mapped above 65K. This was found to be in an unused code fragment from the
original CSCUI development. This unused code is removed.
- During testing, it was noted that only the first communication interface is
brought up during CSC initialization. All others are left in the "Unavail"
state until either an "up" request is made for an interface or an "activate"
request is made for one of their paths. CSC now brings up all configured
interfaces during initialization.
- During testing it was noted that the server returned statuses that were not
properly handled during CSC initialization if the server was also restarted
at the same time. CSC now has code to handle these statuses.
3/20/2003 Changes in CSC 4-1-17
- UCF 94536143 reports an assignment failure when CSC is run with the
QUERYSCRATCH parameter set to TRUE and when all drives in the default ACS are
down. This update adds a check that each LSM recommended by CSC for a
scratch mount contains at least one drive that is up. If all drives in all
LSMs of an ACS are down, then CSC will not recommend using that ACS for the
scratch mount.
2/25/2003 Changes in CSC 4-1-16
- Contact 10824621 reports a misleading console message if the CSC drive
configuration contains drives that are not known to Exec. This update
changes the reason given in that message from the uninformative
"PROCESS_FAILURE" to "UNKNOWN DRIVE."
- Contact 93685833 reports that CSC 4R1 cannot be successfully installed in a
C2 security environment. The guaranteed entry point common bank problems
that caused this are corrected in this update.
- The *CSC CMD keyin for the NCS server is processed using a terminal session
between CSC and the server. This session is not properly closed when CSC is
terminated. This causes *CSC CMD to be unavailable until the NCS server
times out the terminal session. This update changes the CSC termination
timing to allow the *CSC CMD session to be properly closed.
12/12/2002 Changes in CSC 4-1-15
- Contact and UCF 93087274 report problems in configuring CSC to use an L180
library. The L180 library has tape units on panel 0. When CSC was originally
designed, panel 0 could never contain tape units. This property was
incorporated into CSC's internal tests for server availability. This update
changes the server availability checks in CSC to eliminate this dependence.
- An addendum to Contact 93152214 reports an error in the CSC generation to
apply an object module update. The error occurs while producing a merged
report of the changes in the current object module along with all previous
object module update changes. The net result of the error is that the change
report produced by the CSC generation only shows information from the first
object module update. The output master tape from this CSC release is
updated to correct this problem. Note that any CSC object module update
generation based on an output master tape from CSC 4-1-8 through 4-1-14 will
still encounter this problem.
- During internal testing with a new library type, the server returned a
different but equivalent status to the mount request for a volume on a drive
that already contains that same volume but in an unloaded position. The CSC
handling of the alternate status did not contain the appropriate sequence of
actions to recover from this condition so the mount never completed. This
update makes the recovery sequences equivalent for these two mount statuses.
11/20/2002 Changes in CSC 4-1-14
- UCF 92750991 reports that programmatic ejects through the CSC User Interface
do not work after upgrading to CSC 4-1-13 and CSCUI 2-1-13. This is caused
by the cleanup of unused fields in CSCUI 2-1-13. One of the unused fields
that was disallowed has been incorrectly documented as valid since the
inception of the programmatic eject in 1990. Because of the number of
existing programs that may be affected, CSCUI 2-1-14 again allows this unused
field to be passed to CSC.
The error message displayed by CSCUI when there is a problem with passed
fields was also updated. The message now tells if there are extra fields or
missing fields and identifies those fields.
- When CSC receives a dismount request from OS 2200, it inserts a delay before
passing the request to the server. This is designed to allow the rewind and
unload operation to complete so the server will not have to return a mount
failure because the drive is still busy. For some of the newer drive types,
the delay was found to be longer than needed. Those delays have been changed
to a more appropriate value.
10/14/2002 Changes in CSC 4-1-13
- UCF 92182729 reports an abort when either a *CSC QUERY LSM ALL or a *CSC
QUERY POOL ALL command is issued with ACSLS 6. The size has been corrected
for a data area used during CSC initialization. This prevents the memory
corruption that occurs when ACSLS 6 returns a longer response than ACSLS 5.
- UCF 91398634 for CSC 3R5 reports an abort that occurs if a volume report
parameter file contains only blank or empty lines. This abort also occurs in
CSC 4R1. CSC code is changed to return an error status in this situation
rather than attempting to continue processing the volume report.
- While testing the fix for the above problem, it was noted that certain CSCUI
exits allowed the transfer of unexpected data fields. CSCUI code is updated
to pass only the expected data fields. CSC code is updated to eliminate
requests to pass unexpected data fields.
06/18/2002 Changes in CSC 4-1-12
- UCF 90136402 reports problems with CSC builds where the source and
destination are other than tape to tape. The CSC product build routine has
been updated to correct these problems.
- UCF 90052807 against CSC 3R5 reports a communication hang after CSC exhausted
its supply of internal connection entries. Code has been added to time opens
and closes and to manage CSC connection resources in the absence of TSAM
completion notifications.
03/08/2002 Changes in CSC 4-1-11
- During engineering testing of CSC 4-1-10, it was discovered that TSAM
interfaces could not be successfully UPed after CSC had completed
initialization. A correction is included to initialize the variable that
caused this condition.
- While analyzing a CSC 3R5 problem, it was noted that during volume report
transfer, CSC received continuously until all receive buffers were in use
then throttled for a fixed length of time while the buffers were processed.
A change is added to deal with receive buffer status before all buffers are
exhausted.
02/27/2002 Changes in CSC 4-1-10
- A CSC hang resulted when a deadly embrace occurred between a task trying to
resend a timed out request and another task that was beginning to receive
part of the response for the same request. The code in CSC is rearranged to
eliminate the interaction that caused this problem.
- A communication problem caused the server to return a valid ICMP echo
response via an alternate path. This caused CSC to believe that the intended
path was usable when in reality it was only partially functional. CSC now
verifies that path check responses arrive via the expected path.
- UCF 76809751 reports a CSC abort during startup. This occurred when CSC was
testing a communication path and received an ICMP error report instead of the
expected ICMP echo response. CSC now does additional validation of the ICMP
responses during path check.
- UCF 78218407 reports problems when tapes were not successfully dismounted
from DLT 7000 drives. The dismount timeout in CSC was too short for DLT 7000
drives. CSC now has dismount timeouts that are based on drive type. The
timeout for the DLT 7000 is increased to a more reasonable value.
- It was noted during testing that CSC displayed broadcast messages from the
system on which Library Station was running as if they were output from a
*CSC CMD keyin. To prevent this, CSC now displays only server responses
received within 1 minute of issuing the *CSC CMD.
12/20/2001 Changes in CSC 4-1-9
- UCF 19395103 reports that CSC hung during initialization. This happened
because CSC failed to switch paths when CMS returned a "path blocked" status
for the first configured path. The handling of this status is updated to
cause a path switch.
- Corrections are included to prevent CSC hangs under the following conditions:
- CSC is terminated with a large number of requests in progress and while the
active path is experiencing specific error conditions.
- CSC is terminated while the console command session to the NCS server has
failed due to a communication error.
- A *CSC ACTIVATE pathx command is issued while pathx is experiencing certain
communication problems.
- The output of the *CSC STATUS PATH keyin is updated to show the status of a
path that is up via an interface that is down as "unusable" rather than
"available."
12/06/2001 Changes in CSC 4-1-8
- UCF 77195656 reports that CSC does not logoff of the NCS console command
session during CSC termination. This causes delays in establishing the
session when CSC is restarted. CSC now does console command session logoff
during normal CSC termination.
- UCF 76809751 against CSC 3R5 reports a CSC abort condition that also existed
in CSC 4R1. This occurs when CSC receives an error report from ICMP instead
of the expected ICMP echo response. The ICMP response verification in CSC is
updated to ignore all but the expected echo responses.
- The customer reported that CSC experienced several MAX_PATHS failures during
periods of communication problems. This status is generated within CSC to
report a lack of resources condition. Two different code paths were
identified in CSC that failed to release resources when unexpected
communication errors occurred. CSC now releases these RPC program numbers
and TCP connections when these code paths are taken.
- To aid in future problem analysis, additional tracing and operational state
saves have been added to CSC.
- The drive initialization code is rearranged to process groups of drives
through the phases of initialization rather than sequentially initializing
each drive. This greatly decreases the time to initialize all drives.