Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.eso.org/projects/dfs/team/OHS-config-recom.html
Дата изменения: Thu Jan 31 18:55:26 2002
Дата индексирования: Sun Apr 13 22:51:11 2008
Кодировка:

Recommendations for configuring OHS tools

Data Flow System Group

HOME

INDEX

HELP

NEWS

Recommendations and advices for configuring OHS tools

Following checks have to be applied AFTER having installed a new DFS release (according to the latest DFS Installation Guide) and BEFORE running tools such as :

P2PP (java version)
OT (Tcl-tk version)
(DRAFT) OT (java version)
CCSLite environment
BOB

First, be aware of the following Paranal constraints for P2PP configurations:
wudhs : Official installation to create observing blocks. Access to ASTOP. INSIDE the restricted LAN.
pl1 : Offline installation to be used for the visitant astronomers and ScienceOps. Access to PROBS. OUTSIDE the LAN.
pl2 : Offline installation to be used in order to test Instrument Package. This is only used for Paranal Sw Group. Access to PROBS, even if not needed (even in engineering mode, P2PP does access the DB the first time obs runs are to be retrieved: otherwise, no way to use P2PP. Anyway, in engineering mode, access to DB is NOT done when fetching an OB). OUTSIDE the LAN. On pl2, always login as engineering user.

P2PP (java version)

How to modify the DB server access ?

Before running the new P2PP version you installed, under ~visitor, move any previous 'cache' and 'impex' directories to *.old (since it's re-created for each new installation). If not, your old cache and impex directories will be overwritten without warning !
check path for the 'p2pp-db.cf' file: should be under ~/config for previous installations of P2PP (on UT1 and UT2); otherwise, under ~flowmgr/dfs/dataflowJava/config.
login as 'flowmgr' account, and edit the .../config/p2pp-db.cf file : update the line "DATABASE.SERVER" with the IP address and the port number of the new DB server to be accessed.

For P2PP installation WITHIN the restricted Paranal LAN (i.e. for operations) , it should point out ASTOP (otherwise PROBS or a DB server dedicated for testing).
By aware that with P2PP package retrieved from the web, you access by default to the ApplicationServer (i.e. to PROBS).
then, if "encrypted.database.userid" and "encrypted.database.password" need to be updated (no specific need directly related to the new SYBASE configuration, but may be needed for security reason each time the password is to be updated), apply the following procedure :

a) still as 'flowmgr' account, retrieve the file 'p2pp-db.src' from pl1:~cguirao/dfs-patches/p2pp-db.src into the .../config directory.

b) Edit the 'p2pp-db.src' file and modify the following lines according to needs. Be aware, password should be UN-encrypted at this stage.
DATABASE.SERVER "YOUR DATABASE SERVER:YOUR PORT";
ENCRYPTED.DATABASE.PASSWORD "YOUR UN-ENCRYPTED PASSWORD"
Example:
DATABASE.SERVER "YOUR DATABASE "/IP address/:/port number/";
ENCRYPTED.DATABASE.PASSWORD "VerySecret"

Check with Paranal Dataflow team (par-dataflow@eso.org) to get appropriate information about IP, port, username and password. The file is written in PAF format.

c) % cd .../config
% setenv CLASSPATH \
~flowmgr/dfs/dataflowJava/jlib/ohs.jar:$CLASSPATH
% mv p2pp-db.cf p2pp-db.cf.old
% java org.eso.ohs.utilities.EncryptPafFile \
./p2pp-db.src ./p2pp-db.cf

Due to a minor bug, the second argument should not be a bare filename, it needs a directory component, eg ./filename
This command will create the file named as its last argument, which will be the config file.
Usage: java org.eso.ohs.utilities.EncryptPafFile inputfile outputfile
Be aware !!! If it exists, the 'server.cf' file in .../config should be renamed/moved away, if P2PP is to be run NOT via the Application Server : otherwise, P2PP will access automatically the operational DB server PROBS in Garching. This MUST be avoided when working WITHIN the Paranal restricted area and/or when using P2PP for testing purposes ONLY (OUTSIDE the LAN). In principle, Application Server is only needed for external astronomers at home.
check that the latest P2PP script is 'jp2pp' (running the 'which jp2pp' command should return : ~flowmgr/dfs/dataflowJava/bin/jp2pp).
be aware that if the file ~visitor/.p2pp.cf exists, its content supersedes the content of ~visitor/config/site.cf. These two files define access to the following directories : CACHE, IMPEX and INSTRUMENT PACKAGES.
to execute P2PP successfully on HP-UX 11, the Java 1.1.8 needs the calling user (visitor, instmgr, service) to have in his/her environment the variable:
setenv JIT_OPTIONS -nojit
to be set in the .cshrc.local (or .tcshrc.local ?) file for the above mentioned users.
To check current version number of P2PP : go to ~flowmgr/dfs/dataflowJava/lib : there should be either ohs.jar (1 library for several OHS tools), or p2pp.jar (or ot.jar for the next Java-OT).
finally, run : jp2pp &
in case of problem or error message, re-run P2PP in debug mode: jp2pp -debug &
and send corresponding debug messages to SEG opening a DFS Action Remedy Problem Report.

OT (Tcl-tk version)

got to top of page

How to modify the DB server access ?

Under the account 'flowmgr' set the ~flowmgr/dfs/ohs/config/interfaces file as writable (should normally be read-only).
In this 'interfaces' file, add or update lines similar to the following example:
## ASTOP on xxxx.hq.eso.org
## Services:
## query tcp (xxxx)
## master tcp (xxxx)
## Added by Karim, 09-Oct-00, from info by J. Rodriguez, D. Brandt
ASTOP
master tcp ether xxxx xxxx
query tcp ether xxxx xxxx
query tli tcp /dev/tcp \xxxxxxxxxxxxxxx
master tli tcp /dev/tcp \xxxxxxxxxxxxxxx

## ASTOPREP on xxxx.hq.eso.org
## Services:
## query tcp (xxxx)
## master tcp (xxxx)
## Added by Karim, 09-Oct-00, from info by J. Rodriguez
ASTOPREP
master tcp ether xxx.hq.eso.org xxx
query tcp ether xxx.hq.eso.org xxx

These lines declare access to the Paranal Primary AND to the Paranal Replicator machines. Line "master tcp ether ... " and "query tcp ether ..." should contain the machine name and the port number.
Lines "query tli ..." and "master tli ..." are meant for Solaris only (TBC) and defined as follow : ask uss@eso.org
re-set the 'interfaces' file as read only (still under the 'flowmgr' account).
still under 'flowmgr' account, set as writable the file ~flowmgr/dfs/ohs/config/ohs-db.cf.
check this 'ohs-db.cf' file contains the following lines:
# Live databases
OHS_LIVE_OBREP: obrep2
OHS_LIVE_OPC: opc63
OHS_LIVE_SCHED: opc63
If a new DB server is installed, set properly the minimum and maximum OB ID values for this new DB server : add a line similar to:
OHS_SERVER_SEGSRV12: 400000001 500000000
after the part starting by :
# Table of known database servers and related ID ranges.
# We assume that the max number of objects stored in a database
# table is 100,000,000.
#
# OHS_SERVER_server MinID MaxID
#------------------------------------------------------------
OHS_SERVER_PROBS: 0 100000000
OHS_SERVER_OLASLS: 100000001 200000000
OHS_SERVER_ASTOP: 200000001 300000000
OHS_SERVER_TESTSRV: 300000001 400000000
OHS_SERVER_SEGSRV12: 400000001 500000000
OHS_SERVER_USUPLINUX: 500000001 600000000

Min ID and max ID values are to be defined by Paranal Dataflow team (par-dataflow@eso.org).
re-set the 'ohs-db.cf' file as read only (still under the 'flowmgr' account).
If necessary, copy 'otrc' file as '.otrc' from ~flowmgr/dfs/ohs/config to ~visitor. This file gives access to the ot-cache directory used when running OT locally.
path to Instrument Package is found under ~flowmgr/dfs/ohs/instruments : BE AWARE !!! this should be a link to the SAME IP than the one used by P2PP (normally ~instmgr/instruments), otherwise OT will complain when loading queues of OBs.

Then this 'instruments' directory must be g+w.
logout as 'flowmgr', and login as 'visitor'.
setenv DSQUERY as desired DB-server-name (SEGSRV12, ...)
Unsetenv LISP, setenv DISPLAY. If needed, create a new 'ot-cache' directory under ~visitor, to be writable by visitor.
Finally, run : ot &
in case of problem or error message, re-run OT in debug mode: ot -debug Y & (Be aware of the 'Y' !)
and send corresponding debug messages to SEG opening a DFS Action Remedy Problem Report.
If you get an error message similar to "Protocol::getNewID. unknown database server 'XXXX'", it will mean that you didn't properly define your DB server in the ~flowmgr/dfs/ohs/config/ohs-db.cf file or apply the instructions above.
An error message like "Template signature 'xxx' is not supported anymore" means that there is an inconsistency between the TSF files and the index file under the OT .../instruments directory. Notice that these messages can be forgotten if you try only to use OT for testing purposes (the OB should be fetched by BOB anyway, unless ALL his templates are missing). Of course, for operational purposes, TSF definitions have to be consistent between P2PP and OT.
More about OT V1.x configuration files ? go to the USS corresponding page. And finally, further USS instructions for installating OT 1.4.1.

DRAFT : OT (Java version)

got to top of page

Path to the DB server and IP is defined in the
~flowmgr/dfs/dataflowJava/config/ot-db.cf file. To be completed later ...
The $HOME/.ot.cf file is meant for OT V.2.x only (Java OT).

CCSLite environment

got to top of page

before running P2PP (and in order to access CCS and then communicate with BOB), check that the ccs.cf file exists under
~flowmgr/dfs/dataflowJava/config.
If not retrieve it from an operational OHS workstation.
Check if CCSLite is running: ps -ef | grep ccs should display something like:
visitor 6881 1 0 Oct 10 ttyp1 0:00 ccsSHManager -s 20000
visitor 6874 1 0 Oct 10 ttyp1 0:29 ccsScheduler -e wu0ohqs
visitor 6913 1 0 Oct 10 ttyp1 0:00 ccsScan -c 11
if not, check whether a CCS lock files exists under:
/vltdata/ENVIRONMENTS/ccs-environment-name/.ccs-environment-name.lock
example: under /vltdata/ENVIRONMENTS/wu0ohsqs/.wu0ohqs.lock
If so, remove it (this lock file means that a previous session of CCS did not end properly).

Check log files produced by CCS in:

/vlt/ENVIRONMENTS/wg0arc (tail -f rtap_log)
/var/adm/syslog (tail -f syslog.log)
then, under visitor account, run : vccEnvStart -e ccs-environment-name
VLTROOT variable gives indication (only) on the VLT version which is installed.
Check, re-start, stop, ..., a CCS environment ? use the vccEnv tool. "Type" field should be QSEMU (=CCSLite = for P2PP); RTAPENV (FULL CCS=for BOB).
Need more information about CCSLite environment ? click here

BOB

got to top of page

The INS_ROOT environment variable is to be set to ~/OT-P2PP (and not under /data/OT-P2PP as before, because of the merging of the OHS and the OLAS machines) : this is to avoid having visitor account directly accessing to /data which also contains data managed by OLAS).
before running BOB, check that directories: $INS_ROOT/SYSTEM/COMMON/TEMPLATES/TSF and $INS_ROOT/SYSTEM/COMMON/TEMPLATES/SEQ exist.
If not, retrieve them from an operational configuration. Check the UID, GID, and permissions (should be the same as the one from operational configuration, otherwise BOB may also complain).
BE AWARE !!! check that the latest version of the Instrument Packages is ALSO copied in
$INS_ROOT/SYSTEM/COMMON/TEMPLATES/TSF.
This directory should contain the same TSF files as in ~instmgr/instruments (i.e. same TSF files as the ones used by P2PP and OT), otherwise BOB will complain and the corresponding templates will NOT be loaded by BOB.
So, if needed, copy all these TSF files from the current "instruments" directory (the one used by P2PP and OT) to $INS_ROOT/SYSTEM/COMMON/TEMPLATES/TSF.
Then, check that the $INS_ROOT/SYSTEM/COMMON/TEMPLATES/SEQ contains sequencer files. If not it's very easy to re-create (see one existing SEQ file, duplicate it, change the name of the template inside and rename it). For testing purposes only, BOB can run without SEQ files : you will just get a warning message each time an Ob is loaded and a sequencer script is not found.
finally, run bob : bob & (no -debug option available).
TO CHECK IF BOB IS PROPERLY CONFIGURED: open the "Configuration" menu,
OH environment should be name of the OH (OHS) environment.
OS environment should be name of the OS environment.

ps -ef | grep edu on the OH machine should return a process with the OH environment name ps -ef | grep edu on the OS machine should return a process with the OS environment name

check also that $RTAPENV is set with corresponding environment name.
Get a "Could not communicate with OH" error message from BOB when trying to send an OB from P2PP or OT ? see the CCS/InsWs cookbook (courtesy from the VLT Group).