CSWB startup problems

[Bradford Hamilton]

My apologies if this question has been answered; I don't see anything related in the Forum search.

Environment:
VMS V8.3 Alpha
TCPware V5.7-2
CSWB V1.7-13

I have had this problem since VMS V7.3-2, TCPware V5.6-2, and an earlier version of CSWB. The CSWB startup procedure "hangs", without producing a browser window. I can get a "splash" screen using the "-splash" switch, and VMS Mosaic works fine on the same setup.

When I set the debugging option, (VMS_TRACE_FILENAMES) the process hangs reliably at:

open_Jacket: Mapped: /user2/vms$common/cswb/chrome/overlayinfo/navigator/content
/overlays.rdf To: /user2/vms$common/cswb/chrome/overlayinfo/navigator/content/ov
erlays.rdf

The mozilla.log shows *many* polling messages:

28513024[1b08008]: calling PR_Poll [active=0 idle=0]
28513024[1b08008]: PR_Poll timed out

I could attach complete file debugging output, but it's quite large; I'm hoping this is a "known problem", and that I won't need to supply log/debug output.

Anyone else see this behavior?

Thanks,
Brad Hamilton

[Richard Whalen]

It works fine for me with the field test version of TCPware 5.8.

What patches are installed? (VMS and TCPware.) How are you starting CSWB?

[Bradford Hamilton]

Hi Richard,

Thanks for the reply. I was hoping that someone from Process would answer!

I haven't installed V5.8 yet; waiting for Hunter's invite :-)

TCPware patches:
search sys$update:vmsinstal.history 572
| FTP_V572P | 1.0 | 2-APR-2006 | S
| NA | RABBIT | SYSTEM
| SSH_V572P | 1.0 | 2-APR-2006 | S
| NA | RABBIT | SYSTEM
| NETCP_V572P | 1.0 | 21-JUN-2006 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 2.0 | 12-JUL-2006 | F
| NA | RABBIT | SYSTEM
| SSH_V572P | 2.0 | 12-JUL-2006 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 2.1 | 6-AUG-2006 | S
| NA | RABBIT | SYSTEM
| NETCP_V572P | 2.0 | 9-SEP-2006 | S
| NA | RABBIT | SYSTEM
| SSH_V572P | 2.1 | 9-SEP-2006 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 2.2 | 4-NOV-2006 | S
| NA | RABBIT | SYSTEM
| FTP_V572P | 1.0 | 4-NOV-2006 | S
| NA | RABBIT | SYSTEM
| SSH_V572P | 2.1 | 4-NOV-2006 | S
| NA | RABBIT | SYSTEM
| NETCP_V572P | 2.0 | 4-NOV-2006 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 2.2 | 4-NOV-2006 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 3.0 | 4-MAR-2007 | S
| NA | RABBIT | BRAD
| NTP_V572P | 1.1 | 4-MAR-2007 | S
| NA | RABBIT | BRAD
| SMTP_V572P | 2.0 | 17-JUN-2007 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 4.0 | 23-JUN-2007 | S
| NA | RABBIT | SYSTEM
| SSH_V572P | 4.1 | 23-JUN-2007 | S
| NA | RABBIT | SYSTEM
| NETCP_V572P | 2.0 | 25-AUG-2007 | S
| NA | RABBIT | SYSTEM
| DRIVERS_V572P | 5.0 | 25-AUG-2007 | S
| NA | RABBIT | SYSTEM

VMS 8.3 patches:

DEC AXPVMS VMS83A_DRIVER V1.0 Patch Install Val 06-AUG-2007
DEC AXPVMS VMS83A_ACRTL V3.0 Patch Install Val 06-AUG-2007
DEC AXPVMS DNVOSIECO02 V8.3 Patch Install Val 06-AUG-2007
DEC AXPVMS VMS83A_MOUNT96 V3.0 Patch Install Val 29-JUL-2007
DEC AXPVMS VMS83A_BACKUP V3.0 Patch Install Val 29-JUL-2007
DEC AXPVMS DWMOTIF_ECO02 V1.6 Patch Install Val 22-JUN-2007
DEC AXPVMS VMS83A_PASRTL V1.0 Patch Install Val 17-JUN-2007
DEC AXPVMS VMS83A_MOUNT96 V1.0 Patch Install Val 17-JUN-2007
DEC AXPVMS VMS83A_FORRTL V1.0 Patch Install Val 17-JUN-2007
DEC AXPVMS VMS83A_UPDATE V3.0 Patch Install Val 15-JUN-2007
DEC AXPVMS VMS83A_ACMELDAP V2.0 Patch Install Val 20-MAY-2007
DEC AXPVMS DWMOTIF_ECO02 V1.6 Patch Install Val 19-APR-2007
DEC AXPVMS VMS83A_LAN V1.0 Patch Install Val 16-APR-2007
DEC AXPVMS VMS83A_UPDATE V2.0 Patch Install Val 16-APR-2007
DEC AXPVMS VMS83A_ACRTL V1.0 Patch Install Val 16-APR-2007
DEC AXPVMS VMS83A_ERRFMT V1.0 Patch Install Val 16-APR-2007
DEC AXPVMS DWMOTIF_ECO01 V1.6 Patch Install Val 15-APR-2007
DEC AXPVMS VMS83A_BACKUP V2.0 Patch Install Val 14-APR-2007
DEC AXPVMS VMS83A_TZ V1.0 Patch Install Val 07-MAR-2007
DEC AXPVMS VMS83A_UPDATE V1.0 Patch Install Val 03-FEB-2007
DEC AXPVMS DNVOSIECO01 V8.3 Patch Install Val 03-FEB-2007

[Bradford Hamilton]

This is how I start CSWB:

@user2:[vms$common.cswb]mozilla

Thanks,
Brad

[Richard Whalen · Accepted solution]

There aren't any significant differences between the patches that you have installed and TCPware 5.8 as far as getting CSWB to work.

What are the process quotas?

[Bradford Hamilton]

Hi Richard,


Process quotas (for a system with 768Mb of memory and 1.5Gb of pagefile) are as follows:

Maxjobs: 0 Fillm: 4095 Bytlm: 1122030
Maxacctjobs: 0 Shrfillm: 0 Pbytlm: 0
Maxdetach: 0 BIOlm: 4095 JTquota: 4096
Prclm: 10 DIOlm: 12285 WSdef: 81919
Prio: 4 ASTlm: 12385 WSquo: 81919
Queprio: 0 TQElm: 100 WSextent: 393216
CPU: (none) Enqlm: 4096 Pgflquo: 3145728

I've been using the freely-available SHOW_QUOTA.COM to monitor process quotas while starting CSWB. I hardly scratch the surface (1-3%) of any of the quotas monitored.

Thanks,
Brad

[Richard Whalen]

You've got more memory and higher quotas on the system that it works for me on.

Sorrt, but I don't have any more ideas at this point.

[Bradford Hamilton]

Hi Richard,

Thanks for trying. I REMOVE'd and re-INSTALL'ed the kits, this time placing them on the system disk, in the default location. I didn't really think this change would make a difference, and I was right. :-(

Has anyone else seen this problem? It's been frustrating me for a couple of years now.

[Bradford Hamilton]

Some more information:

I originally reported that VMS Mosaic works fine in this environment. This evening, I remembered that there was a version of Netscape for VMS (3.03). I installed that version this evening, and Netscape runs fine, as well.

[Richard Whalen]

I removed TCPware 5.8 from my system and installed TCPware 5.7-2 and DRIVERS_V572p050 and CSWB SWB V1.7-13 starts fine.

What are you using for a display and is the network configured?

[Bradford Hamilton]

Hi Richard,

In all cases, X Window tunneled via SSH from VMS to a Linux laptop. Internal network allows all protocol traffic behind a firewall, where the laptop and VMS reside. I had also tested solely on the VMS machine, without resorting to the network - no luck. To re-iterate, I can also invoke Mozilla with the "-splash" option, which correctly displays a splash screen on my laptop - without displaying the browser.

I'm going to de-install/re-install JAVA next, if I don't have any other bright ideas. Both Mosaic and Netscape (which work fine) don't use JAVA, while Mozilla makes use of JAVA.

[Bradford Hamilton]

I read the documentation more carefully, and realized I didn't have to touch JAVA, so I didn't.

I'll try rebooting the system as a last resort.

[Richard Whalen]

I don't have a system running Linux, so it took me a few days to come up with something that might approximate your configuration:

I ran Microsoft Virtual PC on my Windows XP system and started Ubuntu 6.06 on that. I was able to SSH to a system running TCPware 5.8 and successfully start CSWB.

[Bradford Hamilton]

Thanks Richard,

You've been more than helpful in this thread; unfortunately, I'm still unable to use SWB, even after rebooting the system. I also picked up a freebie monitor from work, and attached it and a mouse and keyboard to my VMS system, and still experienced the same behavior on the "local" display. And all the while, Mosaic and Netscape V3.03 work fine, both "locally" and via SSH to my Linux box (Ubuntu 6.10). I also had the same issue from Windows XP/2000 using the X-window package that comes with the Pathworks CD, and Cygwin. Meanwhile, all other VMS/Motif programs (DECterm, DECClock, File Manager, etc.) all work fine locally, over SSH to Linux, and over the X-Window packages to various Windows boxes.

I guess I'm just going to give up on SWB as a "bad job", and await the next version (or Firefox, when it gets ported) :-)

Thanks again for all your help!

[Willem Grooters]

Did you install the objects, as described in the installation manual? if you, try re-installing the files.

[Bradford Hamilton]

Hi William,

Yes, I neglected to mention that the images get installed at boot time; my understanding was that installing the images saves use of process quotas. I've tried (without success) to get SWB running with and without the installed images.

Just to re-iterate, the SWB process starts, and then "hangs" after a short period of time. I've enabled debugging to see that it hangs while making all sorts of mapping calls to "The Jackets". By reading the comments in the DCL wrapper, one cause of many polling messages (which I see in the logging output) is the use of IPV6, which I don't think I'm using (unless TCPware turns it on by default - I doubt it!). I even attempted to set the logical that told Mozilla that I was using IPV6, but I still couldn't get the procedure to go past the "hanging point".

I'm just thinking now that I have some piece of software or Freeware installed that is preventing SWB from starting up, and the reason that this error is not reproducible is because most folks won't have these "extras" installed on their systems (I think the technical term for this condition is a "corner case"). :-)

Thanks again for your help.

[Willem Grooters]

Some plug-in perhaps? I don't know where to look but no doubt there will be some form of storage that holds the installed plug-in software. It might be useful to get rid of each one by one and see if that helps.

another thing: Have you cehcked user2:[vms$common.cswb.chrome.overlayinfo.navigator.content]overlays.rdf - MIND CASE! Odds are that Mozilla requires PARSE_STYLE=extended, STMLF record format, case sensitivity....

Any idea what port is polled?

[Bradford Hamilton]

Hi Willem,

The .RDF file looks to be "merely" XML "code" that lists a number of Mozilla-related websites.

I ran into a problem recently (documented here in another thread) where *installing* software with PARSE=EXTEND brought me much grief. I've been careful in this instance to install SWB with PARSE=TRAD, but I've been invoking it using PARSE=EXTEND.

As far as the STMLF "format" is concerned, I was a little flummoxed by the release notes; it was unclear to me if I needed to make sure that *all* files were STMLF, or just the ones installed by Mozilla. If the latter is intended, I'm surprised the folks who created the package didn't create code in the installation procedure to do just that!

If anyone can clear up these little mysteries for me, I might have a better idea of how to solve this problem. In the meantime, I'll try a quick de-install/re-install using PARSE=EXTEND, to see if that buys me anything.

[Richard Whalen]

TCPware doesn't implement IPv6, but it needs to tolerate some IPv6 stuff for the version of X-Windows on VMS V8.* to work. It might improve things if you can tell Mozilla to only use IPv4.

Our QA department occaisionally does installs with PARSE=EXTEND and finds interesting things. My argument is that installations should be done from the system account with as few changes to the defaults that the user can tolerate.

[Bradford Hamilton]

OK, I've gotten a little further trying to debug this problem. It seems as though Mozilla is creating a _PARENTLOCK. file in a subdirectory; if Mozilla detects that this file exists, it will not continue to start up. The Mozilla documentation *claims* that this file gets cleaned up on exit, but it does not. If I delete it "maunally" after aborting Mozilla, a new one gets created the next time I run Mozilla.

How do I get Mozilla to *not* create this file?
This is extremely frustrating.

[Steven Schweda]

> [...] if Mozilla detects that this file
> exists, it will not continue to start up.

Really? I see one for each of my profiles:

alp $ dg [...]_par*

Directory ALP$DKA0:[SMS._MOZILLA.DEFAULT.816NKX63_SLT]

_PARENTLOCK.;1 0 19-DEC-2002 17:43:13.05 (RWED,RWED,RE,)

Total of 1 file, 0 blocks.

Directory ALP$DKA0:[SMS._MOZILLA.EBAY.6ZOPQR4H_SLT]

_PARENTLOCK.;1 0 7-AUG-2007 00:55:07.35 (RWED,RWED,RE,)

Total of 1 file, 0 blocks.

Grand total of 2 directories, 2 files, 0 blocks.

And I can pretty much guarantee that I've
restarted CSWB a few times since 19-DEC-2002
and/or 7-AUG-2007, without manually fiddling
either of these files. You have any weird
ownership or security on these files and/or
directories?

[Bradford Hamilton]

I checked file ownership and protections after posting; I own all the files, and have the ability to delete them.

But what do you mean by weird?

In any case, I can easily create a user profile via the "ProfileWizard"; however, any subsequent attempt to do *anything* with Mozilla results in the hanging behavior described previously. It looks as though the only behavior(s) I can get Mozilla to perform are to display the splash screen (@mozilla -splash) and the aforementioned ProfileWizard creation
(@mozilla -ProfileWizard). Any other variation, or combination of variations produces a hang.

DECWindows, DEC Clock, Mosaic, Netscape, etc. all work without issue; it's just Mozilla.

[Bradford Hamilton]

Hi Steven,

Thanks for your response. I did some more research, and now I'm focusing on the file XUL.MSAFL, created in the same directory. This file seems to be a problem, and Googling for XUL.MFL brings up a host of listings that claim that the existence of this file is a problem, and that there are various workarounds to fix the problem (most of the workarounds involve upgrading to Mozilla 1.5, which is not currently an option in our environment. :-))

While I go off tilting at windmills, can I request that folks check their memories for any problems revolving around this file? Also, if folks know where I can get either an "older" version of Mozilla or SWB that may not have this problem, and might work, could you let me know? Again, I'm using the most current version from the hp website (V 1.7.13).

Thanks to all - I think I'm close to getting this problem resolved.

[Steven Schweda]

I don't actually know anything, but here's
what happens when SYSTEM runs CSWB ("Secure
Web Browser V1.7-13 is based on Mozilla
1.7.13") on my main system ("HP TCP/IP
Services for OpenVMS Alpha Version V5.4 -
ECO 6 on a COMPAQ Professional Workstation
XP1000 running OpenVMS V7.3-2"):

ALP $ show defa
SYS$SYSROOT:[SYSMGR]
= SYS$SYSROOT:[SYSMGR]
= SYS$COMMON:[SYSMGR]

ALP $ dg [...]XUL.M*

Directory SYS$COMMON:[SYSMGR._MOZILLA.DEFAULT.6ZOTUZC5_SLT]

XUL.MFASL;1 55 27-JUN-2007 16:44:27.48 (RWED,RWED,RE,)

Total of 1 file, 55 blocks.

ALP $ scswb
%DCL-S-SPAWNED, process SYSTEM_35545 spawned
ALP $
Starting MOZILLA-BIN...

ALP $ dg [...]XUL.M*

Directory SYS$COMMON:[SYSMGR._MOZILLA.DEFAULT.6ZOTUZC5_SLT]

XUL.MFASL;1 1759 17-FEB-2008 10:16:23.24 (RWED,RWED,RE,)

Total of 1 file, 1759 blocks.
ALP $

[Quit the browser manually here.]

Subprocess SYSTEM_35545 has completed
ALP $ dg [...]XUL.M*

Directory SYS$COMMON:[SYSMGR._MOZILLA.DEFAULT.6ZOTUZC5_SLT]

XUL.MFASL;1 1759 17-FEB-2008 10:16:23.24 (RWED,RWED,RE,)

My personal ones are:

alp $ dg [...]XUL.M*

Directory ALP$DKA0:[SMS._MOZILLA.DEFAULT.816NKX63_SLT]

XUL.MFASL;1 47 11-FEB-2008 22:29:44.77 (RWED,RWED,RE,)

Total of 1 file, 47 blocks.

Directory ALP$DKA0:[SMS._MOZILLA.EBAY.6ZOPQR4H_SLT]

XUL.MFASL;1 33 11-FEB-2008 17:20:10.71 (RWED,RWED,RE,)

Total of 1 file, 33 blocks.

Grand total of 2 directories, 2 files, 80 blocks.


So, it appears that one of these things gets
created anew when the browser starts, no one
deletes them at exit, and I seem to have no
problems related to them.

For the record:

ALP $ show symb *cswb
CSWB == "@ SYS$SYSROOT:[CSWB]MOZILLA.COM"
SCSWB == "SPAWN /NOWAIT /NOTIFY /INPUT = NL: CSWB"

[Bradford Hamilton]

Hi Steven,

Thanks, this information helps. My XUL.MSAFL file gets created as a small, blank file initially; this file is *supposed* to be populated with "cached" information once the browser is up and running; since my browser never gets up, the file is destroyed and re-created each time Mozilla attempts to start, always with the same blank information.

Goggling for this phenomenon tells me that I either want to:

Set my preferences to disable this cache, or
Set the file to be a directory, so that Mozilla is "fooled". :-)

I'm working right now to try to set the preferences; (prefs.js file in the same directory) this approach does not seem to be working well. Does anybody know how I can take a file in VMS and "set it's directory bit" using DCL? :-)

In any case, I hope that this information will spur someone to remember some long-forgotten piece of arcana that will allow me to go forward. :-)