Hello all
I started up another VAX to try doing the backups remotely (unable to
connect the tape drive ditrectly to my "original" vax)
I followed the instrictions in the FAQ and it fails miserably...
Anyone has an idea why ?
The 2 servers can see each other and communicate with DECnet (like copying
files, etc).
I don't see what can go wrong and abort the communication link.
Command issued on node HERA:
=======================
$ backup *.com sybile"operateur
macumba"::"task=backup_server"/block=8192/save
NET$SERVER.LOG on node SYBILE
===========================
$ ! Copyright (c) 1995 Digital Equipment Corporation. All rights reserved.
$ IF "" .NES. "" THEN NetServer$COMMAND
$ IF "" .EQS. "" THEN NetServer$VERIFY = 0
$ V = F$VERIFY(NetServer$VERIFY)
--------------------------------------------------------
Connect request received at 15-DEC-2009 10:27:45.69
from remote process HERA::"0=OPERATEUR"
for object "BACKUP_SERVER"
--------------------------------------------------------
$ set noon
$ set rms/network=16
$ allocate mkb400 tapedev
%DCL-I-ALLOC, _SYBILE$MKB400: allocated
$ init mkb400 hera
$ mount/nounload/over:id/block=8192/noassist mkb400
%MOUNT-I-MOUNTED, HERA mounted on _SYBILE$MKB400:
$ convert/fdl=saveset.fdl sys$net mkb400:heradsa0.bck
%CONV-F-OPENIN, error opening HERA::"0=OPERATEUR/
BACKUP_SERVER" as input
-RMS-E-ACC, ACP file access failed
-SYSTEM-F-LINKDISCON, network partner disconnected logical link
$ dismount/nounload mkb400
$ stop/id=0
BTW I always get 2 NET$SERVER.LOG, is it possible that it retries once upon
failure before returning with this error ?
%BACKUP-F-OPENOUT, error opening SYBILE"operateur
password"::"task=backup_server" as output
-RMS-E-CRE, ACP file create failed
-SYSTEM-F-LINKEXIT, network partner exited
I cannot possibly backup to disk, then move that to the tape. The "new"
server has one 300MB drive and one 1GB drive...
Is someone actually doing this successfully ?
Since this was documented I thought it would be easy to do, but apparently
not.
Any help appreciated.
Syltrem
|
|
0
|
|
|
|
Reply
|
syltremzulu (577)
|
12/15/2009 3:40:43 PM |
|
Syltrem wrote:
> $ allocate mkb400 tapedev
> %DCL-I-ALLOC, _SYBILE$MKB400: allocated
> $ init mkb400 hera
> $ mount/nounload/over:id/block=8192/noassist mkb400
> %MOUNT-I-MOUNTED, HERA mounted on _SYBILE$MKB400:
> $ convert/fdl=saveset.fdl sys$net mkb400:heradsa0.bck
> %CONV-F-OPENIN, error opening HERA::"0=OPERATEUR/
> BACKUP_SERVER" as input
Convert wants exclusive access to the file. SYS$NET is already opened by
the decnet task.
Have you tried COPY instead of convert ?
or file copy operayion with BACKUP ?
$BACKUP/IGNORE=INTERLOCK SYS$NET: mkb400:heradsa0.bck
Another possibility would be to pre-mount the tape as a system drive
that anyone can write to. Then the backup operation on HERA would
BACKUP <source> SYBILE::MKB400:heradsa0.bck and let FAL do its magic.
You mght have to adjust block size to 512 if FAL only wants that blocksize.
|
|
0
|
|
|
|
Reply
|
JF
|
12/15/2009 4:05:19 PM
|
|
In article <00e0cbaf$0$6698$c3e8da3@news.astraweb.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:
>
> Another possibility would be to pre-mount the tape as a system drive
> that anyone can write to. Then the backup operation on HERA would
I'm fairly sure you cannot mount/system (or /cluster, or /group)
a tape drive.
|
|
0
|
|
|
|
Reply
|
koehler
|
12/15/2009 5:22:21 PM
|
|
"JF Mezei" <jfmezei.spamnot@vaxination.ca> a �crit dans le message de news:
00e0cbaf$0$6698$c3e8da3@news.astraweb.com...
> Syltrem wrote:
>
>> $ allocate mkb400 tapedev
>> %DCL-I-ALLOC, _SYBILE$MKB400: allocated
>> $ init mkb400 hera
>> $ mount/nounload/over:id/block=8192/noassist mkb400
>> %MOUNT-I-MOUNTED, HERA mounted on _SYBILE$MKB400:
>> $ convert/fdl=saveset.fdl sys$net mkb400:heradsa0.bck
>> %CONV-F-OPENIN, error opening HERA::"0=OPERATEUR/
>> BACKUP_SERVER" as input
>
>
> Convert wants exclusive access to the file. SYS$NET is already opened by
> the decnet task.
>
>
> Have you tried COPY instead of convert ?
>
> or file copy operayion with BACKUP ?
>
> $BACKUP/IGNORE=INTERLOCK SYS$NET: mkb400:heradsa0.bck
>
>
> Another possibility would be to pre-mount the tape as a system drive
> that anyone can write to. Then the backup operation on HERA would
>
> BACKUP <source> SYBILE::MKB400:heradsa0.bck and let FAL do its magic.
> You mght have to adjust block size to 512 if FAL only wants that
> blocksize.
>
>
Hi JF
See
5.42.5 How do I write a BACKUP saveset to a remote tape?
at http://hoffmanlabs.org/vmsfaq/vmsfaq_011.html
That's what I did
The problem is not with SYS$NET per se, but because of
-SYSTEM-F-LINKDISCON, network partner disconnected logical link
I don't know why it disconnects.
Syltrem
|
|
0
|
|
|
|
Reply
|
Syltrem
|
12/15/2009 6:13:37 PM
|
|
Syltrem mentioned on 15-12-2009 19:13:
> "JF Mezei" <jfmezei.spamnot@vaxination.ca> a �crit dans le message de news:
> 00e0cbaf$0$6698$c3e8da3@news.astraweb.com...
>> Syltrem wrote:
>>
>>> $ allocate mkb400 tapedev
>>> %DCL-I-ALLOC, _SYBILE$MKB400: allocated
>>> $ init mkb400 hera
>>> $ mount/nounload/over:id/block=8192/noassist mkb400
>>> %MOUNT-I-MOUNTED, HERA mounted on _SYBILE$MKB400:
>>> $ convert/fdl=saveset.fdl sys$net mkb400:heradsa0.bck
>>> %CONV-F-OPENIN, error opening HERA::"0=OPERATEUR/
>>> BACKUP_SERVER" as input
>>
>> Convert wants exclusive access to the file. SYS$NET is already opened by
>> the decnet task.
>>
>>
>> Have you tried COPY instead of convert ?
>>
>> or file copy operayion with BACKUP ?
>>
>> $BACKUP/IGNORE=INTERLOCK SYS$NET: mkb400:heradsa0.bck
>>
>>
>> Another possibility would be to pre-mount the tape as a system drive
>> that anyone can write to. Then the backup operation on HERA would
>>
>> BACKUP <source> SYBILE::MKB400:heradsa0.bck and let FAL do its magic.
>> You mght have to adjust block size to 512 if FAL only wants that
>> blocksize.
>>
>>
>
> Hi JF
>
> See
> 5.42.5 How do I write a BACKUP saveset to a remote tape?
> at http://hoffmanlabs.org/vmsfaq/vmsfaq_011.html
>
> That's what I did
>
> The problem is not with SYS$NET per se, but because of
> -SYSTEM-F-LINKDISCON, network partner disconnected logical link
>
> I don't know why it disconnects.
>
> Syltrem
This one has always worked for me:
It gives you just one NETSERVER.LOG per session.
$! BACKUP_SERVER.COM
$! Remote tape service via DECnet
$!----------------------------------
$! on tapeless system use:
$!
$! $ BACKUP somefiles -
$! SOMENODE"system password"::"TASK=BACKUP_SERVER" -
$! /BLOCK=16384 /SAVE
$!----------------------------------
$ Set NoOn
$ Set RMS /Network=32 ! 32*512=16384, largest record
$ Alloc MUA3: TAPE
$ Mount /NoAssist /Over=ID /Block=16384 TAPE
$ Convert /Stat SYS$NET TAPE:Remote.BCK /FDL=SYS$INPUT
$DECK
RECORD
FORMAT Fixed
SIZE 16384
$EOD
$ Dismount /Unload TAPE
$ Stop /ID=0
/Wilm
|
|
0
|
|
|
|
Reply
|
Wilm
|
12/15/2009 7:21:00 PM
|
|
Syltrem wrote:
> See
> 5.42.5 How do I write a BACKUP saveset to a remote tape?
> at http://hoffmanlabs.org/vmsfaq/vmsfaq_011.html
If Hoff says it should work, then it should work. He has got a tad more
experience that all of us combined :-)
Is it possible that Backup tried tries to open 2 channels to the outpout
? (hence your 2 log files and the second one failing)
Backup is one of the utilities that has gone trough some fairly radical
changes over the years.
I still think your best bet would be to cluster the 2 nodes, enable
tmscp so that the node 2's tape drive can be used/accessed by node 1.
Another thing you could try is mofify the sys$system:FAL.COM to mount
the tape. Then use a normal decnet.
backup/noinit <source> sybil"shepard pass"::mka500:mysaveset.save/save
Obviously, this would be a test only.
Or perhaps copy FAL.COM to "BACKUP_REMOTE.COM", have it mount the tape
and then invoke FA:.EXE to do the writing. Could that work ?
|
|
0
|
|
|
|
Reply
|
JF
|
12/15/2009 7:33:52 PM
|
|
On Tue, 15 Dec 2009 20:21:00 +0100, Wilm Boerhout <w6.boerhout@planet.nl>
wrote:
> $ Stop /ID=0
Please, excuse this reply that doesn't add any answer to the original
question, but why your example and the one of the original poster end the
process with STOP/ID=0 instead of LOGOUT or just EXIT?
Signed: a learning hobbyist! :-)
G.
|
|
0
|
|
|
|
Reply
|
gerry77
|
12/15/2009 7:52:06 PM
|
|
gerry77@no.spam.mail.com mentioned on 15-12-2009 20:52:
> On Tue, 15 Dec 2009 20:21:00 +0100, Wilm Boerhout <w6.boerhout@planet.nl>
> wrote:
>
>> $ Stop /ID=0
>
> Please, excuse this reply that doesn't add any answer to the original
> question, but why your example and the one of the original poster end the
> process with STOP/ID=0 instead of LOGOUT or just EXIT?
>
> Signed: a learning hobbyist! :-)
Folklore! the STOP/ID is in the FAQ (and may always have been there), so
now it's gospel! As we all know, $STOP/ID=0 is as good as $LOGOUT or
$EOJ here. The idea here is to not let the NETSERVER process linger for
another conenction.
There may some difference in process cleanup or accounting file
processing, but I have not tested this.
/Wilm
|
|
0
|
|
|
|
Reply
|
Wilm
|
12/15/2009 8:05:28 PM
|
|
"Wilm Boerhout" <w6.boerhout@planet.nl> wrote in message
news:4b27e19d$0$14115$703f8584@textnews.kpn.nl...
> This one has always worked for me:
> It gives you just one NETSERVER.LOG per session.
>
> $! BACKUP_SERVER.COM
> $! Remote tape service via DECnet
> $!----------------------------------
> $! on tapeless system use:
> $!
> $! $ BACKUP somefiles -
> $! SOMENODE"system password"::"TASK=BACKUP_SERVER" -
> $! /BLOCK=16384 /SAVE
> $!----------------------------------
> $ Set NoOn
> $ Set RMS /Network=32 ! 32*512=16384, largest record
> $ Alloc MUA3: TAPE
> $ Mount /NoAssist /Over=ID /Block=16384 TAPE
> $ Convert /Stat SYS$NET TAPE:Remote.BCK /FDL=SYS$INPUT
> $DECK
> RECORD
> FORMAT Fixed
> SIZE 16384
> $EOD
> $ Dismount /Unload TAPE
> $ Stop /ID=0
>
>
> /Wilm
Hi
This works for me too.
Haven't yet compared with mine to see the difference, but I succeeded into
backing up to tape !
Many thanks !
Syltrem
|
|
0
|
|
|
|
Reply
|
Syltrem
|
12/16/2009 3:32:46 PM
|
|
In article <4b27ec09$0$14128$703f8584@textnews.kpn.nl>,
Wilm Boerhout <w6.boerhout@planet.nl> wrote:
> gerry77@no.spam.mail.com mentioned on 15-12-2009 20:52:
> > On Tue, 15 Dec 2009 20:21:00 +0100, Wilm Boerhout <w6.boerhout@planet.nl>
> > wrote:
> >
> >> $ Stop /ID=0
> >
> > Please, excuse this reply that doesn't add any answer to the original
> > question, but why your example and the one of the original poster end the
> > process with STOP/ID=0 instead of LOGOUT or just EXIT?
> >
> > Signed: a learning hobbyist! :-)
>
> Folklore! the STOP/ID is in the FAQ (and may always have been there), so
> now it's gospel! As we all know, $STOP/ID=0 is as good as $LOGOUT or
> $EOJ here. The idea here is to not let the NETSERVER process linger for
> another conenction.
>
> There may some difference in process cleanup or accounting file
> processing, but I have not tested this.
>
STOP/ID=0 comes in useful if LOGINOUT.EXE can't be run.
(I had this back on the V3.n era when I was running a VMS upgrade and
someone was logged in when they shouldn't have been.)
--
Paul Sture
|
|
0
|
|
|
|
Reply
|
Paul
|
12/16/2009 9:51:52 PM
|
|
|
9 Replies
710 Views
(page loaded in 0.124 seconds)
|