Use telnet to the alpha, e.g. telnet lpas5. Log in as
TCSMGR. Ask CRB or FJG for the password. The TCS process logs are
under
[tel.DATA.PROCESS], where tel is either INT or WHT.
For example to look at the USER process log for the WHT
TYPE [WHT.DATA.PROCESS]USER.LOG
Note that at the WHT commands issued at the USER> prompt are echoed to the talker syslog on taurus /var/log/talker/talker.info
The TCS login accounts are restricted accounts, which means they
cannot be used to login directly from an Alpha console. If you
need to use the Alpha graphics console to run the TCS, first login
as TCSMGR, then telnet to the the Alpha and then login. For
example, at the INT Alpha, login as TCSMGR then telnet LPAS2
and login as INT_LOGIN. Where the console is a VT220, one can log
in as TCSMGR to see logs, but it won't be possible to start the
TCS displays there.
If neither the TCS DISPLAY nor the TCS USER windows are created
when the TCS is started, then the most likely cause is that the
x-windows creation on the local display is failing. Log files [wht.data.process]user.log
or [wht.data.process]display.log may contain messages
like
Xlib: connection to "_WSA41:" refused by server
Xlib: No protocol specified
%DECW-E-CANT_OPEN_DISPL, Can't open display
Things to check are
One can check creation of a DECterm window from the alpha,
without starting the TCS
$ XDISPLAY
$ CREATE/TERM
Again see more details in confluence.
When the user starts the TCS. The login session disappears. A
detached process called MONITOR runs to start the other TCS
processes. These are started and then MONITOR sends interrupts
(INITAST) to tell each to initialise. When all of the processes
have initialises (warmed up) the the TCS runs. If one of the
processes fails to initialise, there will be a message in
[WHT.DATA.PROCESS]MONITOR.LOG
that
00:00:00 %STATUS-F-TCSNOINI, TCS did not initialise
and MONITOR will close down the TCS processes.
One cause of such a failure appears to have been the autoguider connection. In this case the TCS did not start until the autoguider cable was disconnected from lpas5. If this is suspected, then disconnect the autoguider serial line from the alpha and restart the TCS.
It has also happened on the development machine (lpas3) that the
OUTLOG process did not fully initialise due to a full disk. SHO
DEV D will show disk free space information. The TCS
uses the system disk for storage.
If the TCS appears to start, i.e. it creates the windows which
then disappear soon afterwards, the problem is likely to be in the
CAMAC interface. The problem could be located on the PCI card in
the Alpha, or in the CAMAC controller module in the system crate.
To check whether the CAMAC interface is the problem, login as
TCSMGR and look at the log of the SYNC process:
type [tel.DATA.PROCESS]SYNC.LOG
If the log has the entry Unable to link LAM to AST routine, the
CAMAC interface must be initialised by reloading the CAMAC driver,
which can only be achieved by rebooting the Alpha. It is also a
good idea to power cycle the system crate as well, as follows.
Shut down the Alpha using the SHUT account.
On the CAMAC system crate, switch the branch
coupler (BR CPR 4-1) offline then power off the system crate.
After about 10 seconds, power on the system
crate and wait for the fans to run up.
On the branch coupler, press the branch
initialise switch, then switch to ONLINE.
Boot the Alpha by typing boot at the
>>> prompt.
If the TCS starts, but there is no data in the DISPLAY window,
check that the CAMAC cable is connected correctly. If you need to
reconnect the CAMAC cable, first shutdown the Alpha and power it
off, then switch the branch coupler offline and power off CAMAC.
Check the pins on the connector for damage. Spare cables are kept
in the drawer of the desk in the back right-hand corner of the WHT
computer room. Once the cable is connected securely, power on
CAMAC, initialise the branch coupler and then put it online, then
power on the Alpha which will boot automatically.
If commands cannot be sent to the TCS from the ICS, it may be that the TELD process on the Alpha has died. To restart the TELD process, login as normal and choose the TELD option. On the ICS use the startobssys command to reconnect to the new TELD process.
If the TCS header collection is not working, either the TELD process has died, so the ARCHIVE commands do not work, or else the Unix partition is not mounted on the Alpha. TELD can be restarted as described above. The Unix partition can be remounted by logging in as normal and choosing the NFSMOUNT option.
If the TCS User window stops accepting input, it may be that the
communications have hung
If the TCS Display window stops updating, it may be that the
communications have hung