| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1559.1 | give only disk:[dir] for log files, on SCRATCH disk | CAM::JOHNSON | imagine... sharing all the world | Tue Feb 18 1997 12:30 | 40 |
|
>>BASEstar> sh system
>>%PLI-F-NOSELECT, No WHEN clause selected and no OTHERWISE specified.
this means the conversion did not run...
>>Beginning BASEstar Classic V3.4 Data Base Conversions
>>The BASEstar Classic V3.4 Data Base Conversions can be logged
>>This would provide files with object definitions of the V3.3
>> environment being converted.
>>If requested, these files can be VERY large, and so their
>> location should be a scratch disk, that is NOT the BCC$SYSDATA
>> disk. This way, the conversions can complete, even if the log files
>> use up the entire disk.
>>Do you wish to have object log files created? [Y/N]: y
>>Enter disk:[directory] location for log files: sys$manager:bcc_inst.log
i don't see how much space was on your system disk, but you should REALLY
take the suggestion to pick a scratch disk for the conversion logs...
the DB conversion asked for a disk:[directory] only... each conversion
has a file name...
this should be smarter (ie it should parse the disk/directory out)...
i'll look at fixing that now...
>>BCC$LOGDATA currently translates to:
>> "BCC$LOGDATA" [user] = "SYS$MANAGER:BCC_INST.LOG" (LNM$PROCESS_TABLE)
>>Unable to open the device change log file
>>BASEstar V3.4 Data Base Conversions done...
>>3.4 conversions checked/done...
the conversions did NOT run... due to the invalid location entered above...
you need to set you default to the [.bcc] directory off of you SYS$TEST.
as the SYSTEM account, @execute_v34_db_conversions
and give only a disk:[directory] location...
please use a scratch disk as requested, or you may run out of space
on your system disk.
thankx,
sarah
|
| 1559.2 | conversion log location parser fixed | CAM::JOHNSON | imagine... sharing all the world | Tue Feb 18 1997 19:34 | 9 |
| i have made the parser of the user input for the log area smarter,
so that any file name/type/version info is stripped off, and just the
valid disk:[directory] info is left...
i have re-kitted the .A savesets on cam::bcc$v34_kits_axp and
bcc$v34_kits_vax
with this fix...
|
| 1559.3 | next problem: unable to delete point | VNACO1::KARASEK | Thomas KARASEK @AUI | Thu Feb 20 1997 13:01 | 70 |
| Hi Sarah, thanks so far.
Meanwhile the customer did a new V3.4-installation and added some of his points
manually. When trying to delete one of them, he got the following:
Event 5.10.2 occurred at 20-FEB-1997 12:55:52.14
BCC$$DELETE_POINT (9) - Internal Error: Multiple deallocate attempted on block.
It looks like the point is defined in the point-file, but not in the section.
So he took his original (V3.3) contents ob BCC$SYSDATA and did the conversion
as stated in your previous reply. This worked fine so far, except that the
points still cannot be deleted. He did send me the whole contents of BCC$SYSDATA
for investigation.
The history-file contains the following:
Event 10.1.1 occurred at 20-FEB-1997 12:48:34.86
ILAN$DEVSRV - DCM Device Server Starting.
Event 10.1.3 occurred at 20-FEB-1997 12:48:35.18
ILAN$DEVSRV - DCM Device Server Initializing.
Event 10.86.3 occurred at 20-FEB-1997 12:48:35.68
ILAN$DEVSRV - Section files recreated: ungraceful shutdown or missing files.
Event 10.1.2 occurred at 20-FEB-1997 12:48:36.56
ILAN$DEVSRV - DCM Device Server Running.
Event 1.98.0 occurred at 20-FEB-1997 12:48:37.54
BCC$RPCR - Process up, 0 outstanding requests.
Event 1.98.0 occurred at 20-FEB-1997 12:48:47.11
BCC$MONITOR - No site specific command file present.
Event 1.99.0 occurred at 20-FEB-1997 12:49:29.21
Multiple deallocate attempted on block.
Event 5.10.2 occurred at 20-FEB-1997 12:49:29.32
The following message pertains to point APMSS_FOX .
Event 5.10.2 occurred at 20-FEB-1997 12:49:29.32
BCC$$DELETE_POINT (9) - Internal Error: Multiple deallocate attempted on block.
Event 2.2.5 occurred at 20-FEB-1997 12:49:29.48
Point APMSS_FOX not fully deleted.
Multiple deallocate attempted on block.
Event 2.1.5 occurred at 20-FEB-1997 12:55:19.23
User ABFS02::APMSDEV created point APMSS_XXX.
Event 1.99.0 occurred at 20-FEB-1997 12:55:52.00
Multiple deallocate attempted on block.
Event 5.10.2 occurred at 20-FEB-1997 12:55:52.08
The following message pertains to point APMSS_XXX .
Event 5.10.2 occurred at 20-FEB-1997 12:55:52.14
BCC$$DELETE_POINT (9) - Internal Error: Multiple deallocate attempted on block.
Event 2.2.5 occurred at 20-FEB-1997 12:55:52.25
Point APMSS_XXX not fully deleted.
Multiple deallocate attempted on block
...
As you can see, he did delete the section-files before starting BASEstar again.
There's no DAS currently running, and the points he created have were just
/WRITEABLE /FORMAT=WORD.
Since he got the same problem either with converting his old database or
creating points after a brand-new installation, I'm quite lost.
- regards, -Tom.
|
| 1559.4 | looks like alignment bug in name section | CAM::JOHNSON | imagine... sharing all the world | Thu Feb 20 1997 15:46 | 74 |
| Hi,
>>Meanwhile the customer did a new V3.4-installation and added some of his points
>>manually. When trying to delete one of them, he got the following:
was this another V3.3 upgrade, or a new install on a new system?
if a new install, was it a stand-alone system, or part of a domain, and if a
domain, what is the gns/nd configuration?
>>Event 5.10.2 occurred at 20-FEB-1997 12:55:52.14
>>BCC$$DELETE_POINT (9) - Internal Error:
>>Multiple deallocate attempted on block.
this is a problem with the name, not the point object itself...
>>It looks like the point is defined in the point-file, but not in the section.
what make you say this... did you search bcc$sysdata:*.dat for the name?
if so, what do you get?
>>So he took his original (V3.3) contents ob BCC$SYSDATA and did the conversion
>>as stated in your previous reply. This worked fine so far, except that the
>>points still cannot be deleted. He did send me the whole contents of BCC$SYSDATA
>>for investigation.
now i'm confused...
- how was the V3.4 installation done (clean install/upgrade/etc)
- what is the configuration
- how/when was the point created (during V3.3, then upgraded, after
a v3.4 upgrade, after a clean v3.4 install)
- what command is used to create the point
- what does a BS SET NAME/READ/TYPE=5/scope=3 POINT_NAME give you
- VERY IMPORTANT>>> are these points defined as global names??
Event 1.99.0 occurred at 20-FEB-1997 12:49:29.21
Multiple deallocate attempted on block.
Event 5.10.2 occurred at 20-FEB-1997 12:49:29.32
The following message pertains to point APMSS_FOX .
Event 5.10.2 occurred at 20-FEB-1997 12:49:29.32
BCC$$DELETE_POINT (9) - Internal Error: Multiple deallocate attempted on block.
Event 2.2.5 occurred at 20-FEB-1997 12:49:29.48
Point APMSS_FOX not fully deleted.
Multiple deallocate attempted on block.
Event 2.1.5 occurred at 20-FEB-1997 12:55:19.23
User ABFS02::APMSDEV created point APMSS_XXX.
Event 1.99.0 occurred at 20-FEB-1997 12:55:52.00
Multiple deallocate attempted on block.
Event 5.10.2 occurred at 20-FEB-1997 12:55:52.08
The following message pertains to point APMSS_XXX .
Event 5.10.2 occurred at 20-FEB-1997 12:55:52.14
BCC$$DELETE_POINT (9) - Internal Error: Multiple deallocate attempted on block.
Event 2.2.5 occurred at 20-FEB-1997 12:55:52.25
Point APMSS_XXX not fully deleted.
Multiple deallocate attempted on block
>>As you can see, he did delete the section-files before starting BASEstar again.
>>There's no DAS currently running, and the points he created have were just
>>/WRITEABLE /FORMAT=WORD.
definitely not /scope=global??
>>Since he got the same problem either with converting his old database or
>>creating points after a brand-new installation, I'm quite lost.
so he did a clean install AND a data base conversion??
i DO believe there is a bug in basestar, and i am looking at fixing it now...
i'll have to get you a new BCC$SHR to try... i'll re-read these notes,
but if it doesn't say, is this VAX or Alpha??
please have them do a BS DUMP NAMC/FU/OUT=namc.dat
and send me namc.dat... i am especially curious as to the decnet id of the
system...
in the future, can you please submit customer problems to cam::bstar_mail,
along with the template that was sent to all the basestar_support members?
(if you need a new tempate, just let me know)
|
| 1559.5 | BS SET PARAM BCC$MAX_NAMES_CACHED 0, restart and delete point | CAM::JOHNSON | imagine... sharing all the world | Thu Feb 20 1997 16:03 | 21 |
| hi again...
i'm having trouble reproducing, so i'm going thru the debugger to
see if i can verify what i think the problem is...
i THINK they can temporarily solve this by doing the following
(please do this AFTER you did the namc dump i asked for in the last note/mail)
BS SET PARAM BCC$MAX_NAMES_CACHED 0
restart basestar
try and delete the point again...
if this works, it will save others a great deal of agony until we get a fix
out.
please let me know how this goes as soon as you can.
thankx,
sarah
|
| 1559.6 | work-around works, only problem on certain systems | CAM::JOHNSON | imagine... sharing all the world | Thu Feb 20 1997 18:00 | 23 |
| here i am back again...
quite the obscure problem your customer has found...
setting the bcc$max_names_cached to 0 will definitely solve the problem
for now... there may be a slight performance degradation, as all name
functions will take a file i/o hit, but it shouldn't be too bad...
i will be fixing this and releasing a new bcc$shr for both vax and alpha in the
next few days...
we will also write a STARS article on this... but i still need to verify
with you that the problem is as 'controlled' as i think it is...
this should happen ONLY on a system whose SYSGEN SCSSYSTEMID has a hex
value where the low-order byte = 2
please do post the scssystemid of the customer system...
i want to verify that i have a handle on the full problem before
implementing the fix.
thankx,
sarah
|
| 1559.7 | Results ... | VNACO1::KARASEK | Thomas KARASEK @AUI | Fri Feb 21 1997 11:06 | 39 |
| Hi Sarah !
Thanks so far (though I didn't post it the official way).
First of all:
- yes you are right, SCSSYSTEMID = 40962 (Hex = 0000A002)
- BS SET PARAM BCC$MAX_NAMES_CACHED 0 resolves that issue for
any points created afterwards. However, you still cannot delete
any of the previously created points.
- NAMC.DAT is on the way, I'll send it per mail asap.
>> - how was the V3.4 installation done (clean install/upgrade/etc)
Both - After a 'clean install' and manually creating the points
the problem was discovered. Thereafter we did restore BCC$SYSDATA with
the previously saved V3.3-files, and called the conversion-routine
EXECUTE_V34_DB_CONVERSIONS.COM - with the same symtom.
>> - what is the configuration
Only about 50 logical points, 2 of them being structures, the rest
should be format:word.
>> - how/when was the point created (during V3.3, then upgraded, after
>> a v3.4 upgrade, after a clean v3.4 install)
$ BS CREAT POINT /WRITE /FORMAT=WORD <pointname>
both after a clean V3.4-install, as well as after upgrading the
V3.3-database again.
>> - what does a BS SET NAME/READ/TYPE=5/scope=3 POINT_NAME give you
SYSID SYSNAM TYPE SCOPE OBJECT ID OBJECT NAME
------ ------ ---------------- --------- ----------- ------------
-24574 ABFS02 ( 5)Point (1)Local 43409792 GUE_TEST
>> - VERY IMPORTANT>>> are these points defined as global names??
No !
- Thanks & regards, Tom.
|
| 1559.8 | moved to support channels | CAM::JOHNSON | imagine... sharing all the world | Fri Feb 21 1997 11:23 | 0
|