You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Gregory Youngblood <gr...@netio.org> on 2005/10/20 15:25:36 UTC
db_recover fatal error, is repo lost?
Bad BDB errors - how can I recover from this?
Continuing from my previous request for help on recovering my
repository, I used VMware and went to an older install where I built
subversion 1.0.9 with db 4.1.25. This matches the installed software
from my repository that I recovered from backup. I have two repos.
Using this method, I was able to svnadmin dump one of them succesfully.
But, not the second, larger one. And, of course, not the one I want back
the most.
First I tried svnadmin dump. It produced a very small dump file (195
bytes) with 0 revisions. svnadmin recover didn't seem to do anything
either. So, I tried db_recover. This is what I got:
db_recover: Finding last valid log LSN: file: 444 offset 658689
db_recover: Recovery starting from [1][28]
db_recover: DB_LOGC->get: invalid log record header
db_recover: PANIC: Input/output error
db_recover: fatal region error detected; run recovery
db_recover: fatal region error detected; run recovery
db_recover: DB_ENV->open: DB_RUNRECOVERY: Fatal error, run database
recovery
Something tells me it's really bad when running db_recovery to be told
to run database recovery on a fatal error.
Is this repo lost to all time? Or is there something I can do to get it
back?
Thanks,
Greg
Re: db_recover fatal error, is repo lost?
Posted by Gregory Youngblood <gr...@netio.org>.
On Fri, 2005-10-21 at 13:41 +0200, Ryan Schmidt wrote:
> On Oct 21, 2005, at 08:45, Gregory Youngblood wrote:
>
> > Well - I thought I did. These _were_ my backups. I just didn't have
> > an "svnadmin dump" backup.
> >
> > I kept going through my backups and finally found a backup set that
> > worked, and successfully dumped the BDB based repo and loaded it
> > into an FSFS repo. I think it's complete -- if not, I lost 1 or 2
> > months of changes at the most, and the changes during that time
> > would have been trivial anyway. So I'm happy. Happy that I kept
> > more than just 2 or 3 backup sets. It was the 5th or 6th set I
> > found that I was finally able to recover.
>
>
> In the past, some of the Subversion developers who monitor this list
> have offered, in situations like these, to accept a copy of the
> broken repository to see if they can fix it. I was under the
> impression that these attempts are generally successful... though I
> haven't seen such an offer on the list in a month or two. But maybe
> this message will prompt one?
>
Since I did manage to get my repo back, I don't need the recovery
effort. If it does prompt such an offer, I'd rather it go to someone
that wasn't as fortunate as I was to locate and recover from backups.
Thx,
Greg
Re: db_recover fatal error, is repo lost?
Posted by Ryan Schmidt <su...@ryandesign.com>.
On Oct 21, 2005, at 08:45, Gregory Youngblood wrote:
> Well - I thought I did. These _were_ my backups. I just didn't have
> an "svnadmin dump" backup.
>
> I kept going through my backups and finally found a backup set that
> worked, and successfully dumped the BDB based repo and loaded it
> into an FSFS repo. I think it's complete -- if not, I lost 1 or 2
> months of changes at the most, and the changes during that time
> would have been trivial anyway. So I'm happy. Happy that I kept
> more than just 2 or 3 backup sets. It was the 5th or 6th set I
> found that I was finally able to recover.
In the past, some of the Subversion developers who monitor this list
have offered, in situations like these, to accept a copy of the
broken repository to see if they can fix it. I was under the
impression that these attempts are generally successful... though I
haven't seen such an offer on the list in a month or two. But maybe
this message will prompt one?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: db_recover fatal error, is repo lost?
Posted by Gregory Youngblood <gr...@netio.org>.
Well - I thought I did. These _were_ my backups. I just didn't have an
"svnadmin dump" backup.
I kept going through my backups and finally found a backup set that
worked, and successfully dumped the BDB based repo and loaded it into an
FSFS repo. I think it's complete -- if not, I lost 1 or 2 months of
changes at the most, and the changes during that time would have been
trivial anyway. So I'm happy. Happy that I kept more than just 2 or 3
backup sets. It was the 5th or 6th set I found that I was finally able
to recover.
Thanks,
Greg
On Thu, 2005-10-20 at 17:25 +0100, Rachael Russell wrote:
> You do have backups right??!!!!! If not, you may be stuck. Everytime i
> had a fatal BDB error, I couldn't get the repository back no matter
> what, so switched to using fsfs and rebuilt with weekly full + nightly
> incremental backups, and not much was lost.
>
> Backups are a life saver...! The scripts in the pragmatic programmer
> book are perfect, and i'm sure there are hundreds of variants around
> on this list.
>
>
>
> On 10/20/05, Gregory Youngblood <gr...@netio.org> wrote:
>
> Bad BDB errors - how can I recover from this?
>
> Continuing from my previous request for help on recovering my
> repository, I used VMware and went to an older install where I
> built subversion 1.0.9 with db 4.1.25. This matches the
> installed software from my repository that I recovered from
> backup. I have two repos.
>
> Using this method, I was able to svnadmin dump one of them
> succesfully. But, not the second, larger one. And, of course,
> not the one I want back the most.
>
> First I tried svnadmin dump. It produced a very small dump
> file (195 bytes) with 0 revisions. svnadmin recover didn't
> seem to do anything either. So, I tried db_recover. This is
> what I got:
>
> db_recover: Finding last valid log LSN: file: 444 offset
> 658689
> db_recover: Recovery starting from [1][28]
> db_recover: DB_LOGC->get: invalid log record header
> db_recover: PANIC: Input/output error
> db_recover: fatal region error detected; run recovery
> db_recover: fatal region error detected; run recovery
> db_recover: DB_ENV->open: DB_RUNRECOVERY: Fatal error, run
> database recovery
>
> Something tells me it's really bad when running db_recovery to
> be told to run database recovery on a fatal error.
>
> Is this repo lost to all time? Or is there something I can do
> to get it back?
>
> Thanks,
> Greg
>
>
Re: db_recover fatal error, is repo lost?
Posted by Rachael Russell <ra...@gmail.com>.
You do have backups right??!!!!! If not, you may be stuck. Everytime i had a
fatal BDB error, I couldn't get the repository back no matter what, so
switched to using fsfs and rebuilt with weekly full + nightly incremental
backups, and not much was lost.
Backups are a life saver...! The scripts in the pragmatic programmer book
are perfect, and i'm sure there are hundreds of variants around on this
list.
On 10/20/05, Gregory Youngblood <gr...@netio.org> wrote:
>
> Bad BDB errors - how can I recover from this?
>
> Continuing from my previous request for help on recovering my repository,
> I used VMware and went to an older install where I built subversion 1.0.9with db
> 4.1.25. This matches the installed software from my repository that I
> recovered from backup. I have two repos.
>
> Using this method, I was able to svnadmin dump one of them succesfully.
> But, not the second, larger one. And, of course, not the one I want back the
> most.
>
> First I tried svnadmin dump. It produced a very small dump file (195
> bytes) with 0 revisions. svnadmin recover didn't seem to do anything either.
> So, I tried db_recover. This is what I got:
>
> db_recover: Finding last valid log LSN: file: 444 offset 658689
> db_recover: Recovery starting from [1][28]
> db_recover: DB_LOGC->get: invalid log record header
> db_recover: PANIC: Input/output error
> db_recover: fatal region error detected; run recovery
> db_recover: fatal region error detected; run recovery
> db_recover: DB_ENV->open: DB_RUNRECOVERY: Fatal error, run database
> recovery
>
> Something tells me it's really bad when running db_recovery to be told to
> run database recovery on a fatal error.
>
> Is this repo lost to all time? Or is there something I can do to get it
> back?
>
> Thanks,
> Greg