%STDRV-F-FATALERR, fatal error reading startup database

[Willem Grooters]

Stating platform and VMS version would be handy...

[quote]
$open/share/error=comp_error stdrv$product1_0 STARTUP$STARTUP_VMS
$comp_error:
$stdrv$say f$fao("%STDRV-E-OPENFAIL, failure opening component file !UB, !AS", s
tdrv$startup_file, stdrv$comp_file)
%STDRV-E-OPENFAIL, failure opening component file 0, STARTUP$STARTUP_VMS
[/quote]

Conclusion: Something is wrong with file "STARTUP$STARTUP_VMS". I have no Alpha at hand to see what it should refer to (as a logical of a file specification).
As Bob Gezelter already mentioned: check your file structure. If your system disk is ODS-5, check case of directory- and filenames. IIRC, I once had a problem on boot from an ODS-5 disk where one of the required directories had not been named in uppercase.

[Hakan Zanderau ( Anders]

The logical STARTUP$STARTUP_VMS points to SYS$STARTUP:VMS$VMS.DAT

Check if the file exist or if you can read it.

Hakan Zanderau

[Willem Grooters]

And what is shown by

$ SHO LOG/FULL SYS$STARTUP
$ SHO DEV/FULL SYS$STARTUP

The latter to find out if it's a ODS-2 or ODS-5 structured disk. As stated before, non-uppercase directory names _could_ cause trouble - here, or before this point.

[Clark Powell]

Sorry about the delay in answering; this is a test system and I was involved in production issues. Priority on this problem is low as I don't need to fix this problem; I will simply use our disaster recovery tapes and create a new one. On the other hand, it would be useful to know why this system failed so profoundly, out of the blue. A test system can be abused and I do abuse it but I can think of nothing that would have caused this problem. Here's a little bit more data. If we can't figure the problem out with this data then I would like to put it aside. If it happens again I will have more input as to how it go this way.

One thing I don't understand about what's going on here is why the logical, sys$system is not defined yet it works as
@sys$system:startup.com There may aspect of OpenVMS at work here that I don't understand.
Also why is the default directory
DSA10:[SYSEXE]?
$ dir
%DIRECT-E-OPENIN, error opening DSA10:[SYSEXE]*.*;* as input
-RMS-E-DNF, directory not found
-SYSTEM-W-NOSUCHFILE, no such file
$ sho def
DSA10:[SYSEXE]
%DCL-I-INVDEF, DSA10:[SYSEXE] does not exist

disk is ODS2
Volume Status: ODS-2, subject to mount verification, write-through caching
enabled.


$ show log sys$system:
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name SYS$SYSTEM:
$ sho log startup$startup_list
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name STARTUP$STARTUP_LIST
$ sho log startup$startup_list
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name STARTUP$STARTUP_LIST
$ dir
%DIRECT-E-OPENIN, error opening DSA10:[SYSEXE]*.*;* as input
-RMS-E-DNF, directory not found
-SYSTEM-W-NOSUCHFILE, no such file
$
$ sho log SYS$STARTUP
sho log SYS$STARTUP
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name SYS$STARTUP
$ show log STARTUP$STARTUP_VMS
show log STARTUP$STARTUP_VMS
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name STARTUP$STARTUP_VMS
$ show log sys$system
%SYSTEM-W-IVLOGTAB, invalid logical name table
%SHOW-S-NOTRAN, no translation for logical name SYS$SYSTEM

$ @SYS$SYSTEM:STARTUP.COM
%STDRV-I-STARTUP, OpenVMS startup begun at 8-JAN-2009 09:12:09.18
$else
$endif
$stdrv$phase_pos = 0
$stdrv$phase_len = 12
$stdrv$mode_pos = 12
$stdrv$mode_len = 1
$stdrv$file_pos = 13
$stdrv$file_len = 79
$stdrv$param_pos = 93
$stdrv$enable_flag = "E"
$stdrv$disable_flag = "D"
$if p1.nes."FULL"
$then stdrv$vms_only = stdrv$true
$else stdrv$vms_only = stdrv$false
$endif
$stdrv$param = "Initial Parameters"
$stdrv$invoc = f$integer(f$trnlnm("stdrv$invoc","lnm$startup_table",,"EXECUTIVE"
))
$stdrv$invoc = stdrv$invoc + 1
$define/exec/nolog/table=lnm$startup_table stdrv$invoc 1
$stdrv$phase_name = "stdrv$phase" + f$string(stdrv$invoc)
$stdrv$product_name = "stdrv$product" + f$string(stdrv$invoc)
$stdrv$save_layered = f$trnlnm("startup$startup_layered","lnm$system_table")
$stdrv$save_num = -1
$stdrv$startup_file = 0
$next_open_file:
$stdrv$comp_file = f$trnlnm("startup$startup_list","lnm$system_table",stdrv$star
tup_file)
$if stdrv$comp_file .eqs. "" then goto count_comp_files
$if stdrv$comp_file .eqs. "STARTUP$STARTUP_LAYERED" then stdrv$save_num = stdrv$
startup_file
$if stdrv$verbose then stdrv$say f$fao("%STDRV-I-OPENCOMP, opening component fil
e !UB, !AS", stdrv$startup_file, stdrv$comp_file)
$open/share/error=comp_error stdrv$product1_0 STARTUP$STARTUP_VMS
$comp_error:
$stdrv$say f$fao("%STDRV-E-OPENFAIL, failure opening component file !UB, !AS", s
tdrv$startup_file, stdrv$comp_file)
%STDRV-E-OPENFAIL, failure opening component file 0, STARTUP$STARTUP_VMS
$stdrv$say f$fao("%STDRV-I-CONTINUING, continuing with partial file list")
%STDRV-I-CONTINUING, continuing with partial file list
$count_comp_files:
$stdrv$comp_files = stdrv$startup_file - 1
$if f$trnlnm(stdrv$phase_name,"lnm$process_table").nes."" then $ close stdrv$pha
se1
$open/share=read stdrv$phase1 STARTUP$PHASES
$get_phase:
$read/end=done_all_phases stdrv$phase1 stdrv$current_phase
$if stdrv$verbose then -
stdrv$say f$fao("!/%STDRV-I-STRTPHAS, starting phase !AS at !%D",stdrv$current_p
hase,0)
$define/nolog startup$phase_name INITIAL
$stdrv$num_batch_jobs = 0
$stdrv$startup_file = 0
$goto first_file
$first_file:
$if stdrv$verbose then stdrv$say f$fao("%STDRV-I-INDXFILE, using component file
!UB", stdrv$startup_file)
$stdrv$first = stdrv$true
$if stdrv$startup_file .eq. stdrv$save_num .and. -
stdrv$save_layered .nes. f$trnlnm("startup$startup_layered","lnm$system_table")
$endif
$goto get_product
$get_product:
$if stdrv$first
$then
$read/end=next_file/error=read_error/nolock -
/index=0/key=INITIAL stdrv$product1_0 stdrv$product_info
$read_error:
$stdrv$error_code == $status
$if stdrv$error_code.eq.%X000182B2
$endif
$stdrv$say "%STDRV-F-FATALERR, fatal error reading startup database"
%STDRV-F-FATALERR, fatal error reading startup database
$exit stdrv$error_code
%DCL-W-UNDFIL, file has not been opened by DCL - check logical name
$

[Hakan Zanderau ( Anders · Accepted solution]

Did you check the file SYS$STARTUP:VMS$VMS.DAT ?

[VMS$COMMON.SYS$STARTUP]VMS$VMS.DAT

You get the first errormessage because the file can't be opened

Hakan

[Robert Gezelter]

Clark,

I do not have my references handy. I would have to check why SYS$SYSTEM seems to be undefined yet appears to work at that point.

Basically, if one gets control BEFORE STARTUP.COM executes, the system is not completely functional. There is a bare minimum functionality, and STARTUP.COM is crafted to not work outside of these limits. As the startup sequence progresses, more and more of the system is usable.

By the time one gets to LP* phases, almost all of the system is usable.

I do not see an answer to what is the status of the [VMS$COMMON.SYS$STARTUP] directory. At this point in the life of the system, I generally DO NOT use logical names, I explicitly enumerate the path [That may be an acquired quirk from the dark ages, but it is depoendable].

- Bob Gezelter, http://www.rlgsc.com

[Clark Powell]

OK, thanks Bob, that explains a lot. And thanks, Hakan, I did find that SYS$STARTUP:VMS$VMS.DAT is missing and when I copied one from a backup I got a plethora of new errors. I don't think that this disk is salvageable but I do think that I should be thoughtful to see if I can recreate the problem because it may be unrecoverable in any case. I am using some beta distributions so I need to go through the motions of installing these and see if this is the source of the problem. It difficult because the problem only makes booting impossible; the system keeps running until one tries to reboot. It sure does emphasize the golden rule of new software and that is, keep a shadow copy of the system disk for an orderly retreat. So I'm going to focus on figuring out what caused this and warning any developers if necessary. Following this will be the log which I present for curiosity only.

%EIA0, Link state change: UP
%EIB0, Link state change: UP
%SHADOW-I-VOLPROC, DSA10: is offline. Volume Processing in progress.

%SHADOW-I-VOLPROC, DSA10: has completed volume processing.

%STDRV-I-STARTUP, OpenVMS startup begun at 8-JAN-2009 10:50:17.40
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$INITIAL-050_VMS.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$INITIAL-050_LIB.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$INITIAL-050_CONFIGURE.COM does not exi
st.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_CSP.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_SHADOW_SERVER.COM does not
exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$DEVICE_STARTUP.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:IPC$STARTUP.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_VMS.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_ERRFMT.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_CACHE_SERVER.COM does not e
xist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_OPCOM.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_AUDIT_SERVER.COM does not e
xist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_JOBCTL.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_LMF.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_SECURITY_SERVER.COM does no
t exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$CONFIG-050_ACME_SERVER.COM does not ex
ist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$BASEENVIRON-050_VMS.COM does not exist
.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$BASEENVIRON-050_SMISERVER.COM does not
exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$BASEENVIRON-050_LIB.COM does not exist
.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:DECDTM$STARTUP.COM does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:LICENSE_CHECK.EXE does not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$BASEENVIRON-050_INDICT_SERVER.COM does
not exist.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$LPBEGIN-050_STARTUP.COM does not exist
.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$LPBEGIN-050_SMHANDLER.COM does not exi
st.
%STDRV-E-NOSUCHFILE, File SYS$STARTUP:VMS$END-050_COORDINATED.COM does not exist
.
SYSTEM job terminated at 8-JAN-2009 10:50:20.26

Accounting information:
Buffered I/O count: 173 Peak working set size: 5184
Direct I/O count: 80 Peak virtual size: 183296
Page faults: 203 Mounted volumes: 0
Charged CPU time: 0 00:00:00.04 Elapsed time: 0 00:00:03.07

[Robert Gezelter]

Clark,

Please compare a directory listing of [VMS$COMMON.SYS$STARTUP] and [SYSn.SYSMGR] on the "damaged" system and its undamaged BACKUP.

You are getting error messages indicating that the various component files invoked by STARTUP.COM are missing.

Without data, my guess would be that somebody deleted the contents of this directory. Note that SYS$STARTUP (the logical name) includes both the SYS$SPECIFIC and SYS$COMMON contents of BOTH [SYS$STARTUP] and [SYSMGR] (you can see this on any properly running OpenVMS system by using SHOW LOGICAL). This logical name search sequence is what allows LAT$STARTUP.COM to be in SYS$COMMON:[SYS$STARTUP] (which is actually [VMS$COMMON.SYS$STARTUP]) and LAT$SYSTARTUP.COM to be in [SYSMGR] (actually, any of the four directories referenced above; normally clusterwide LAT$SYSTARTUP.COM is in SYS$COMMON:[SYSMGR] if it is node-specific, it is in SYS$SPECIFIC:[SYSMGR]).

As I mentioned, check ALL of the files. Working one at a time will just slow progress.

- Bob Gezelter, http://www.rlgsc.com

[EdgarZamora_1]

It is possible that you think files are missing when they're actually not. If I'm not mistaken you are using logical names when you look for the files. Your logicals are hosed. Use the actual directory paths... i.e. instead of looking for SYS$STARTUP:VMS$VMS.DAT look for [VMS$COMMON.SYS$STARTUP]VMS$VMS.DAT.

What do

SYSGEN> SHOW STARTUP
SYSGEN> SHOW STARTUP_P1 ... Pn show?

[Clark Powell]

No, files are missing; I'm checking the actual directorys and not logicals. I think that it's possible that a PRODUCT REMOVE command did this but I will have to try to reproduce the problem to find out. Thanks, everyone for the help.