imported restore fails - partially
A user on Informix 7.31UC5 / Solaris with onbar/ISM, Sun T3 RAID and Veritas Volume Manager reported that an imported restore only partly worked: about two-thirds of the dbspaces restored, while the rest stayed down (ND) instead of entering recovery (NR), with assertion failures and later rsclose_phr errors, even though onbar -r -v against the backup set was clean. IBM support blamed cross-device link errors, but re-creating volumes/symlinks didn't help. A respondent suggested examining the assertion failure (af) file and its page dump, and testing with cooked files to see whether the OS exposes broken hardware/RAID/Veritas problems. No resolution is recorded in the thread.
Auto-generated by DrWatson from the posts below — may be imperfect; read the full thread.
I am encountering a problem with an imported restore, using onbar/ISM,
INFORMIX 7.31UC5,
Solaris 6, and VERITAS volume manager. Running onbar -r -v against the
backup set yields
no errors, so the problem must be on the target system.
The symptoms are this...
Some spaces restore, about 2/3 of them. For the others, the restore
progresses as expected,
but the spaces are never marked as being in a recovery state (NR).
Spaces remain down
(ND) even though the onbar status and debug files indicate that it is
reading the tape and
writing to the spaces. We were getting assertion failures at the end
of the 'restore' process
for the failed spaces, but are now getting rsclose_phr errors.
INFORMIX tech support has
told us that this is being caused by cross-device link errors -
however, we have deleted
and re-created all the volumes and links, and the process still fails.
We are now in the
process of rebuilding the target system from the ground up, but I
don't expect this to solve
the problem.
Any ideas? Suggestions?
Randy Goshorn
Northrup-Grumman Corp.
What are the errors now???
> and re-created all the volumes and links, and the process still fails.
assume that you do not create 2 sym links to the same device???
Superboer.
On 8 apr, 16:41, rgosho...@comcast.net wrote:
> I am encountering a problem with an imported restore, using onbar/ISM,
> INFORMIX 7.31UC5,
> Solaris 6, and VERITAS volume manager. Running onbar -r -v against the
> backup set yields
> no errors, so the problem must be on the target system.
>
> The symptoms are this...
>
> Some spaces restore, about 2/3 of them. For the others, the restore
> progresses as expected,
> but the spaces are never marked as being in a recovery state (NR).
> Spaces remain down
> (ND) even though the onbar status and debug files indicate that it is
> reading the tape and
> writing to the spaces. We were getting assertion failures at the end
> of the 'restore' process
> for the failed spaces, but are now getting rsclose_phr errors.
> INFORMIX tech support has
> told us that this is being caused by cross-device link errors -
> however, we have deleted
> and re-created all the volumes and links, and the process still fails.
> We are now in the
> process of rebuilding the target system from the ground up, but I
> don't expect this to solve
> the problem.
>
> Any ideas? Suggestions?
>
> Randy Goshorn
> Northrup-Grumman Corp.
On Apr 10, 11:05 am, "Superboer" <superbo...@t-online.de> wrote:
> What are the errors now???
>
> > and re-created all the volumes and links, and the process still fails.
>
> assume that you do not create 2 sym links to the same device???
>
> Superboer.
>
> On 8 apr, 16:41, rgosho...@comcast.net wrote:
>
>
>
> > I am encountering a problem with an imported restore, using onbar/ISM,
> > INFORMIX 7.31UC5,
> > Solaris 6, and VERITAS volume manager. Running onbar -r -v against the
> > backup set yields
> > no errors, so the problem must be on the target system.
>
> > The symptoms are this...
>
> > Some spaces restore, about 2/3 of them. For the others, the restore
> > progresses as expected,
> > but the spaces are never marked as being in a recovery state (NR).
> > Spaces remain down
> > (ND) even though the onbar status and debug files indicate that it is
> > reading the tape and
> > writing to the spaces. We were getting assertion failures at the end
> > of the 'restore' process
> > for the failed spaces, but are now getting rsclose_phr errors.
> > INFORMIX tech support has
> > told us that this is being caused by cross-device link errors -
> > however, we have deleted
> > and re-created all the volumes and links, and the process still fails.
> > We are now in the
> > process of rebuilding the target system from the ground up, but I
> > don't expect this to solve
> > the problem.
>
> > Any ideas? Suggestions?
>
> > Randy Goshorn
> > Northrup-Grumman Corp.- Hide quoted text -
>
> - Show quoted text -
...still getting the rsclose_phr assertion error.
I've checked and double-checked to make sure there are not two links
pointing to the same volume, or any sort of 'overlap'. Does not appear
to be the case.
Is it possible the the Sun T3 RAIDS or the VERITAS volume manager
could have introduced an error internally?
On 11 apr, 00:38, rgosho...@comcast.net wrote:
> On Apr 10, 11:05 am, "Superboer" <superbo...@t-online.de> wrote:
>
>
>
>
>
>
>
> > What are the errors now???
>
> > > and re-created all the volumes and links, and the process still fails.
>
> > assume that you do not create 2 sym links to the same device???
>
> > Superboer.
>
> > On 8 apr, 16:41, rgosho...@comcast.net wrote:
>
> > > I am encountering a problem with an imported restore, using onbar/ISM,
> > > INFORMIX 7.31UC5,
> > > Solaris 6, and VERITAS volume manager. Running onbar -r -v against the
> > > backup set yields
> > > no errors, so the problem must be on the target system.
>
> > > The symptoms are this...
>
> > > Some spaces restore, about 2/3 of them. For the others, the restore
> > > progresses as expected,
> > > but the spaces are never marked as being in a recovery state (NR).
> > > Spaces remain down
> > > (ND) even though the onbar status and debug files indicate that it is
> > > reading the tape and
> > > writing to the spaces. We were getting assertion failures at the end
> > > of the 'restore' process
> > > for the failed spaces, but are now getting rsclose_phr errors.
> > > INFORMIX tech support has
> > > told us that this is being caused by cross-device link errors -
> > > however, we have deleted
> > > and re-created all the volumes and links, and the process still fails.
> > > We are now in the
> > > process of rebuilding the target system from the ground up, but I
> > > don't expect this to solve
> > > the problem.
>
> > > Any ideas? Suggestions?
>
> > > Randy Goshorn
> > > Northrup-Grumman Corp.- Hide quoted text -
>
> > - Show quoted text -
>
> ...still getting the rsclose_phr assertion error.
>
> I've checked and double-checked to make sure there are not two links
> pointing to the same volume, or any sort of 'overlap'. Does not appear
> to be the case.
>
> Is it possible the the Sun T3 RAIDS or the VERITAS volume manager
> could have introduced an error internally?
rsclose_phr assertion comes with an af file and may be/hopefully with
a hex dump of a page it does not like.
based on this info one can have a look on disk to see what is going
on.
It is not entirely impossible that your sun has memory problems or the
raid config is broken or
veritas has a problem......
What did TS had to say about this???
Superboer.
TS says it's a cross-device link error. We have not found anything
in the volume name/'link pointer configuration to supprt this
conjecture,
thus we are rebuilding the target system. i.e. reconfiiguring the T3
RAID/VERITAS configuration as if it were newly out-of-the-box.
Superboer wrote:
> On 11 apr, 00:38, rgosho...@comcast.net wrote:
> > On Apr 10, 11:05 am, "Superboer" <superbo...@t-online.de> wrote:
> >
> >
> >
> >
> >
> >
> >
> > > What are the errors now???
> >
> > > > and re-created all the volumes and links, and the process still fails.
> >
> > > assume that you do not create 2 sym links to the same device???
> >
> > > Superboer.
> >
> > > On 8 apr, 16:41, rgosho...@comcast.net wrote:
> >
> > > > I am encountering a problem with an imported restore, using onbar/ISM,
> > > > INFORMIX 7.31UC5,
> > > > Solaris 6, and VERITAS volume manager. Running onbar -r -v against the
> > > > backup set yields
> > > > no errors, so the problem must be on the target system.
> >
> > > > The symptoms are this...
> >
> > > > Some spaces restore, about 2/3 of them. For the others, the restore
> > > > progresses as expected,
> > > > but the spaces are never marked as being in a recovery state (NR).
> > > > Spaces remain down
> > > > (ND) even though the onbar status and debug files indicate that it is
> > > > reading the tape and
> > > > writing to the spaces. We were getting assertion failures at the end
> > > > of the 'restore' process
> > > > for the failed spaces, but are now getting rsclose_phr errors.
> > > > INFORMIX tech support has
> > > > told us that this is being caused by cross-device link errors -
> > > > however, we have deleted
> > > > and re-created all the volumes and links, and the process still fails.
> > > > We are now in the
> > > > process of rebuilding the target system from the ground up, but I
> > > > don't expect this to solve
> > > > the problem.
> >
> > > > Any ideas? Suggestions?
> >
> > > > Randy Goshorn
> > > > Northrup-Grumman Corp.- Hide quoted text -
> >
> > > - Show quoted text -
> >
> > ...still getting the rsclose_phr assertion error.
> >
> > I've checked and double-checked to make sure there are not two links
> > pointing to the same volume, or any sort of 'overlap'. Does not appear
> > to be the case.
> >
> > Is it possible the the Sun T3 RAIDS or the VERITAS volume manager
> > could have introduced an error internally?
>
> rsclose_phr assertion comes with an af file and may be/hopefully with
> a hex dump of a page it does not like.
> based on this info one can have a look on disk to see what is going
> on.
> It is not entirely impossible that your sun has memory problems or the
> raid config is broken or
> veritas has a problem......
>
> What did TS had to say about this???
>
> Superboer.
On 11 apr, 20:41, rgosho...@comcast.net wrote: > TS says it's a cross-device link error. We have not found anything > in the volume name/'link pointer configuration to supprt this > conjecture, you could also try .........yuk...........barf...... sorry ........ cooked files. BE PATIENT... it will take a lot longer.... may be the os will tell you then if something is broken... if that is the case then you have broken hardware... Superboer.