Webes v5 failes to start on some cluster nodes?

[James Cordill]

I have installed webes v5 on to my 4 node v7.3-2 alpha cluster which has multiple system disks. The install went fine, and two of the nodes are running fine, the other two, one of which is the installation node refuses to start webes. Here is the log output from one of the systems, hope someone can point me to what's wrong.

$ desta start
Starting WCCProxy:
%RUN-S-PROC_ID, identification of created process is 222D4B4B
%RUN-S-PROC_ID, identification of created process is 222BFF51
Director in startup phase 2.
Director in startup phase 2.
The Director process has not finished startup.
It may not have started correctly.
Please refer to _$1$DGA5:[WEBES.HP.NODES.OVMS3.SVCTOOLS.SPECIFIC.WEBES.LOGS]DEST
A_DIR.LOG; for more details

$ type/tail $1$DGA5:[WEBES.HP.NODES.OVMS3.SVCTOOLS.SPECIFIC.WEBES.LOGS]DES
TA_DIR.LOG;
$ VER = F$VERIFY( 0 )
JAVA "-DSvctools.Home=/SVCTOOLS_HOME" "-noverify" "-Xdynclassgc" "-Xmx512M" com.
compaq.svctools.desta.core.DESTAController
___.
CRIPPLED FROM ERROR on November 12, 2007 7:25:31 AM CST (0.916 sec elapsed)
Service: Socket failed to start.
Current Thread[Thread-1: com.compaq.svctools.desta.core.DESTAProcess_Thr for
:5769190360027935812,5,main]
___.
CRIPPLED FROM ERROR on November 12, 2007 7:25:31 AM CST (0.931 sec elapsed)
Target constructor of DESTAComponent com.compaq.svctools.desta.core.Comm
onDirector threw an exception in its thread, did not start.
Current Thread[Thread-1: com.compaq.svctools.desta.core.DESTAProcess_Thr for
:5769190360027935812,5,main]
EXCEPTION Director unable to create listener thread.
One or more CommMechanismListener Services failed to start
at com.compaq.svctools.desta.core.DESTADirector.(DESTADirector.jav
a:79)
at com.compaq.svctools.desta.core.CommonDirector.(CommonDirector.j
ava:49)
at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstruct
orAccessorImpl.java:39)
at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingC
onstructorAccessorImpl.java:27)
at java.lang.reflect.Constructor.newInstance(Constructor.java:494)
at com.compaq.svctools.desta.core.DESTAProcess_Thr.startup(DESTAProcess_
Thr.java:105)
at com.compaq.svctools.desta.core.DESTAProcess.run(DESTAProcess.java:178
)
at java.lang.Thread.run(Thread.java:595)
Exception in thread "Thread-1: com.compaq.svctools.desta.core.DESTAProcess_Thr f
or :5769190360027935812" java.lang.NullPointerException
at com.compaq.svctools.desta.core.DESTAController.stopDESTAProcess(DESTA
Controller.java:1042)
at com.compaq.svctools.desta.core.DESTAController.stopDESTAProcess(DESTA
Controller.java:1130)
at com.compaq.svctools.desta.core.DESTAProcess_Thr.startup(DESTAProcess_
Thr.java:115)
at com.compaq.svctools.desta.core.DESTAProcess.run(DESTAProcess.java:178
)
at java.lang.Thread.run(Thread.java:595)
___.
FATAL ERROR on November 12, 2007 7:35:31 AM CST (600.63 sec elapsed)
The Director failed to start - Shutting down DESTA now

Current Thread[main,5,main]

[Wim Van den Wyngaert]

I read "unable to create listener thread.".

Could it be that an TCP port is still open ?
Are all part stopped ? My version has WXCCProxy, DESTA director and CA* process names. Stop/id if any are left.

I have a daily restart procedure that stop webes, kills everything and restarts it twice (sometimes the first start fails). But I have a very old version.

Wim

[Jan van den Boogaard]

I had lots of troubles too in getting Webes 5.0 to work on a cluster:

When you read the install guide, section 5.1.8 tells that its perfectly oke to install webes on a non-system target disk. Webes will be installed on all nodes that "see" the target disk.

I specified the target disk in ANSWER_FILE.TXT as follows:
inst_path_input=MYDISK:[000000]
This works, webes will be installed in MYDISK:[HP...] directory tree.

Section 5.3.1.1 states that you must run WCCPROXYCLUSTER.COM and WEBESCLUSTER.COM to install webes on NEW nodes in the cluster.

WEBESCLUSTER.COM firstly has a typing error in this section:

----
$
$ ! Deleting the webesInstalled.txt File
$
$ ! if (webes44status .eqs. 1)
$ ! then
$ ! say "WEBES is Installed"
$ ! if
F$SEARCH("SVCTOOLS_HOME:[000000.specific.desta.data]webesInstalled.txt") .eqs.
""
$ ! then
$ ! DELETE/noconfirm
SVCTOOLS_HOME:[000000.specific.desta.data]webesInstalled.txt
$ ! endif
$ endif
$ type/page NLA0:
$
-----

this gives an if-then-else inconsistency error. You can easily mend it yourself.

Second, webescluster.com has a annoying habit of defining SVCTOOLS* logicals that still point to the system disk instead of the non-systemdisk that webes was installed on. This is a bit harder to mend, but still can be done.

Thirdly, there is a typing error in DESTA.COM :
"-Xm512M" should be "-Xmx512M".
Correct this, or you might get errors: "OpenVMS Could not create the Java virtual machine."


greetz Jan