You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@subversion.apache.org by Ritesh Nadhani <ri...@webyog.com> on 2004/11/10 07:38:11 UTC
Re: DB corruption of first day of subversion
Hello,
I had posted my issued on a wrong list but Jani Averach was kind enough to
tell me a solution. Find below complete information on the issue.
>>>>>>>>>>
Hi, I replied to your message to the svn-dev list.
I don't know if you are a subscriber of that list or not, so I am
replying privately to you also.
BTW, where do you find that issue-list address? This address is meant
to be for internal bugtracking purposeses, general issues/bug reports
should go to user-svn.
So my reply to your post was:
Any changes that your F: drive is a network share (mapped networkd
drive)?
If so, do NOT run SVN repository on shared drive. Put it on the
local disk, and everything should be fine.
http://svnbook.red-bean.com/svnbook-1.1/ch05s02.html
BR, Jani
On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
> Hello,
>
> We plan to move all our repository from CVS to subversion.
>
> We are using WinXP with SP2. I was successfuly able to create create a
> repository and even was able to import my test project to the repository.
> Our server is running as a standalone svnserve daemon listening on 3690
> port. It is on 512Kbps dsl line (bandwith is pretty costly out here).
>
> One of our developers was also able to download the test project (around 5
> MB with documenation and all) to his home machine. He did some
> modification
> to the code and checked in. All this process were OK.
>
> Now, suddenly looks like the whole berkely DB is corrupt. Whenever we are
> now trying to create a new project in the repository - we are getting an
> error: Berkely DB error while allocating new node ID (getting 'next-key' )
> for filesystem f:\svnrepos/db, DB_NOTFOUND: No matching key/data pair
> found.
>
> If I try to checkout a project, we get error:
>
> Reference to non-existent node 'd.0.1' in filesystem 'F:/svnrepos/db'
>
> What should I do? Is SVN known to have bugs with Windows XP SP2.
>
> Regards,
> Ritesh
>
>>>>>>>>
F: is just a partition on my hard-disk. I have 4 partition on the box.
c:\windows - XP Files
d:DATA - General data files
e:softwares - all softwares are installed out here instead of \program
files\ . Makes it easy to manage.
f:\repos - I am using it only for Subversion repository so that backup are
easy to do.
This drive is not network mapped. I mean you cannot access it from any
machine in the networked. The only folder on this machine that can be
accessed by other people is the SHareDocs which is present in C;\. Disk
management shows it as:
Type: Local disk
Filesystem: NTFS.
Am I missing something?
Regards,
Ritesh
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by "Alex R. Mosteo" <al...@mosteo.com>.
Jani Averbach wrote:
> On 2004-11-11 17:42+0100, Alex R. Mosteo wrote:
>
>>Maybe. My own stress test cases regularly make fail the BDB repository
>>but have no effect in the FSFS repository.
>
>
> Can you share your recipe? It would be superb if you could provide a
> recipe to corrupt BDB, which isn't caused just by something on your
> system.
I posted it some time ago, you can search for a thread with "stall" or
"stalls" in the topic. If I remember well, at least another person was
having the same problem (BTW it was not corruption but deadlock).
As I said, I was using a precompiled RPM and the BDB in use was 4.2.
Looking at the BDB website I saw two patches for the last version at
that time, but I didn't manage to determine if these were applied in my
installed libraries.
Another relevant aspect of my problem is that I used that installation
for months until I hit the problem, which only was evident when
committing *lots* of files (i.e the probability of it happening with few
files is much smaller, not nil). Any Mandrake user with a default
installation could have this problem and not be aware of it (that's
another reason I posted my test script).
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Jani Averbach <ja...@jaa.iki.fi>.
On 2004-11-11 17:42+0100, Alex R. Mosteo wrote:
>
> Maybe. My own stress test cases regularly make fail the BDB repository
> but have no effect in the FSFS repository.
Can you share your recipe? It would be superb if you could provide a
recipe to corrupt BDB, which isn't caused just by something on your
system.
BR, Jani
--
Jani Averbach
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by "Alex R. Mosteo" <al...@mosteo.com>.
Mark wrote:
> Nearly all "corruption" issues with BDB are permissions issues that are
> easily avoided by following the documentation's instructions and easily
> corrected. Of the rest, almost all is caused by running outdated versions of
> BDB, esp 4.1 on OSX, and running BDB databases on network filesystems.
Well, you have just listed some caveats which may go unnoticed for some
time by a newcomer. I was using 4.2 on linux on a local drive. I don't
compile subversion myself, I use the precompiled .rpm from subversion
homepage... so I suppose these have the dependencies right.
> Maybe FSFS is more highly stable, but it's got nowhere the exposure of the
> BDB FS, so judging the backends by search hits on this list for corruption
> isn't a super good idea.
Maybe. My own stress test cases regularly make fail the BDB repository
but have no effect in the FSFS repository. Obviously my choice is
between using a proven wrong BDB or a FSFS with unknown bugs. Which
isn't really a choice since I'm obliged to use FSFS and hope that if
some FSFS nasty bug arises my backup is not too old.
Of course YMMV, but both my personal experience and my following of this
list for several months hint to avoid BDB. And now we have a new 4.3
release of BDB which has still less exposure than FSF.
So: In any case, backup often ;)
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello,
----- Original Message -----
From: "Alex R. Mosteo" <al...@mosteo.com>
To: <us...@subversion.tigris.org>
Sent: Thursday, November 11, 2004 4:44 PM
Subject: Re: DB corruption of first day of subversion
> Ritesh Nadhani wrote:
>> Hello,
>>
>> ----- Original Message ----- From: "Alex R. Mosteo"
>> <al...@mosteo.com>
>>>
>>> - Avoid BerkeleyDB
>>
>>
>> Even I am use BerkelyDB. Do you say I should use FSFS?
>
> I have had problems with BDB. I haven't had problems with FSFS. It's only
> my experience, but if I remember well, I still haven't heard a single
> complaint of corruption from someone using FSFS. From BDB, instead, you
> can find aplenty.
>
>>
>>> - Try some humongous imports/exports/commits/updates with lots of files.
>>
>>
>> The funny part is that it was doing ok with 2-3 commits for around 1 day.
>> After that I did one full checkout and one full commit from my home
>> machine to the server and then it started failing everytime. The checkout
>> was of around 10MB (it included doxygen documentation for the whole
>> source code). Do you think that might be the reason and I move to FSFS?
>
> "Failing everytime". What that means? In my case, the repository seemed
> locked forever and coulnd't be recovered. What I mean is that there
> weren't error messages after the one in the first failure, nor progressive
> corruption... A failed commit and then nothing in any operation.
>
In my case it was like. I did a full import to the subversion repository.
Went home checked out everything. Made some changes. Commited it back and
whoa - i start getting all sort of invlid node problems in BDB. I went back
to office and did a local commit/export and same sort of node invalidity.
>>> - Make backups :)
>>
>> For this to happen, my subversion has to work for atleast one week :)
>
> That's the same period I use.
>
> My bottom line: if you haven't any special reason to use BDB, give FSFS a
> try.
I think I will do that but I will take backups everyday till I am sure, its
working great!!!
>
Thanks for your advice.
Regards,
Ritesh
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
RE: DB corruption of first day of subversion
Posted by Mark <ma...@msdhub.com>.
Nearly all "corruption" issues with BDB are permissions issues that are
easily avoided by following the documentation's instructions and easily
corrected. Of the rest, almost all is caused by running outdated versions of
BDB, esp 4.1 on OSX, and running BDB databases on network filesystems.
Maybe FSFS is more highly stable, but it's got nowhere the exposure of the
BDB FS, so judging the backends by search hits on this list for corruption
isn't a super good idea.
Mark
-----Original Message-----
From: news [mailto:news@sea.gmane.org] On Behalf Of Alex R. Mosteo
Sent: Thursday, November 11, 2004 4:15 AM
To: users@subversion.tigris.org
Subject: Re: DB corruption of first day of subversion
Ritesh Nadhani wrote:
I have had problems with BDB. I haven't had problems with FSFS. It's
only my experience, but if I remember well, I still haven't heard a
single complaint of corruption from someone using FSFS. From BDB,
instead, you can find aplenty.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by "Alex R. Mosteo" <al...@mosteo.com>.
Ritesh Nadhani wrote:
> Hello,
>
> ----- Original Message ----- From: "Alex R. Mosteo" <al...@mosteo.com>
>>
>> - Avoid BerkeleyDB
>
>
> Even I am use BerkelyDB. Do you say I should use FSFS?
I have had problems with BDB. I haven't had problems with FSFS. It's
only my experience, but if I remember well, I still haven't heard a
single complaint of corruption from someone using FSFS. From BDB,
instead, you can find aplenty.
>
>> - Try some humongous imports/exports/commits/updates with lots of files.
>
>
> The funny part is that it was doing ok with 2-3 commits for around 1
> day. After that I did one full checkout and one full commit from my home
> machine to the server and then it started failing everytime. The
> checkout was of around 10MB (it included doxygen documentation for the
> whole source code). Do you think that might be the reason and I move to
> FSFS?
"Failing everytime". What that means? In my case, the repository seemed
locked forever and coulnd't be recovered. What I mean is that there
weren't error messages after the one in the first failure, nor
progressive corruption... A failed commit and then nothing in any operation.
>> - Make backups :)
>
> For this to happen, my subversion has to work for atleast one week :)
That's the same period I use.
My bottom line: if you haven't any special reason to use BDB, give FSFS
a try.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello,
----- Original Message -----
From: "Alex R. Mosteo" <al...@mosteo.com>
To: <us...@subversion.tigris.org>
Sent: Thursday, November 11, 2004 3:12 PM
Subject: Re: DB corruption of first day of subversion
> Ritesh Nadhani wrote:
>> Hello again,
>>
>> ----- Original Message ----- From: "Jani Averbach" <ja...@jaa.iki.fi>
>> To: "Ritesh Nadhani" <ri...@webyog.com>
>> Cc: <us...@subversion.tigris.org>
>> Sent: Wednesday, November 10, 2004 11:25 PM
>> Subject: Re: DB corruption of first day of subversion
>>
My advices from some months using Subversion, for when you arrive to try
it:
>
> - Avoid BerkeleyDB
Even I am use BerkelyDB. Do you say I should use FSFS?
> - Try some humongous imports/exports/commits/updates with lots of files.
The funny part is that it was doing ok with 2-3 commits for around 1 day.
After that I did one full checkout and one full commit from my home machine
to the server and then it started failing everytime. The checkout was of
around 10MB (it included doxygen documentation for the whole source code).
Do you think that might be the reason and I move to FSFS?
> - Make backups :)
For this to happen, my subversion has to work for atleast one week :)
>
>
Regards,
Ritesh
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by "Alex R. Mosteo" <al...@mosteo.com>.
Ritesh Nadhani wrote:
> Hello again,
>
> ----- Original Message ----- From: "Jani Averbach" <ja...@jaa.iki.fi>
> To: "Ritesh Nadhani" <ri...@webyog.com>
> Cc: <us...@subversion.tigris.org>
> Sent: Wednesday, November 10, 2004 11:25 PM
> Subject: Re: DB corruption of first day of subversion
>
>
>> On 2004-11-10 15:54+0530, Ritesh Nadhani wrote:
>>
>>>
>>> Now, everytime I run the check, the TEST column goes from 1 to 100% and
>>> and
>>> then it again goes back to 2% and it just stops. None of the keys work
>>> i.e.
>>> Esc,(C)ancel and the two other keys. I need to reboot my machine.
>>>
>>> This is the final report that is being shown after the freeze:
>>>
>>> Pentium 4 2400 Mhz
>>> | Pass 1%
>>> L1 Cache 8K 19670 MB/s
>>> | Test 2%
>>> L2 Cache 512K 16782 MB/s
>>> | Test [Address Test, own address, no cache]
>>> Memory 511M 88 MB/s
>>> | Testing 80K 511 M 511M
>>> Chipset /* amazingly its blank but I have a geniune Intel
>>> motherboard */
>>>
>>> Walltime Cached RsvdMem MemMap Cache ECC
>>> Test Pass Errors ECC Errs
>>>
>>> 0:0012 511M 832K e820-std off off std 0 0 0
>>>
>>> Is there something wrong?
>
>
> I also think so :(
>
>>
>> There is something wrong, memtest should run in the loop, and it
>> should not freeze. If this is a reason of something in your HW or
>> something in memtest86, I can't tell.
>
>
> Well, its stopping in my box but works on other :(
>
>>
>> As someone said, you could try http://www.memtest.org/, and see what
>> it says (I didn't know that it exists).
>
>
> I tried this one also. Its better coz it recognises my motherboard to be
> original Intel make. But the problem is same. After reaching:
>
> Pass: 2%
> Test: 100%
>
> It just freezes. I cant rebood nor the configuration windows pops up. I
> think I will just check up with my hardware guy and lets see what he says.
> Hardware failure sucks :( I badly wanted to try subversion before we start
> the actual project.
My advices from some months using Subversion, for when you arrive to try it:
- Avoid BerkeleyDB
- Try some humongous imports/exports/commits/updates with lots of files.
- Make backups :)
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello again,
----- Original Message -----
From: "Jani Averbach" <ja...@jaa.iki.fi>
To: "Ritesh Nadhani" <ri...@webyog.com>
Cc: <us...@subversion.tigris.org>
Sent: Wednesday, November 10, 2004 11:25 PM
Subject: Re: DB corruption of first day of subversion
> On 2004-11-10 15:54+0530, Ritesh Nadhani wrote:
>>
>> Now, everytime I run the check, the TEST column goes from 1 to 100% and
>> and
>> then it again goes back to 2% and it just stops. None of the keys work
>> i.e.
>> Esc,(C)ancel and the two other keys. I need to reboot my machine.
>>
>> This is the final report that is being shown after the freeze:
>>
>> Pentium 4 2400 Mhz
>> | Pass 1%
>> L1 Cache 8K 19670 MB/s
>> | Test 2%
>> L2 Cache 512K 16782 MB/s
>> | Test [Address Test, own address, no cache]
>> Memory 511M 88 MB/s
>> | Testing 80K 511 M 511M
>> Chipset /* amazingly its blank but I have a geniune Intel motherboard */
>>
>> Walltime Cached RsvdMem MemMap Cache ECC
>> Test Pass Errors ECC Errs
>>
>> 0:0012 511M 832K e820-std off off std 0 0 0
>>
>> Is there something wrong?
I also think so :(
>
> There is something wrong, memtest should run in the loop, and it
> should not freeze. If this is a reason of something in your HW or
> something in memtest86, I can't tell.
Well, its stopping in my box but works on other :(
>
> As someone said, you could try http://www.memtest.org/, and see what
> it says (I didn't know that it exists).
I tried this one also. Its better coz it recognises my motherboard to be
original Intel make. But the problem is same. After reaching:
Pass: 2%
Test: 100%
It just freezes. I cant rebood nor the configuration windows pops up. I
think I will just check up with my hardware guy and lets see what he says.
Hardware failure sucks :( I badly wanted to try subversion before we start
the actual project.
>
> BR, Jani
>
Regards,
Ritesh
> --
> Jani Averbach
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Jani Averbach <ja...@jaa.iki.fi>.
On 2004-11-10 15:54+0530, Ritesh Nadhani wrote:
>
> Now, everytime I run the check, the TEST column goes from 1 to 100% and and
> then it again goes back to 2% and it just stops. None of the keys work i.e.
> Esc,(C)ancel and the two other keys. I need to reboot my machine.
>
> This is the final report that is being shown after the freeze:
>
> Pentium 4 2400 Mhz
> | Pass 1%
> L1 Cache 8K 19670 MB/s
> | Test 2%
> L2 Cache 512K 16782 MB/s
> | Test [Address Test, own address, no cache]
> Memory 511M 88 MB/s
> | Testing 80K 511 M 511M
> Chipset /* amazingly its blank but I have a geniune Intel motherboard */
>
> Walltime Cached RsvdMem MemMap Cache ECC
> Test Pass Errors ECC Errs
>
> 0:0012 511M 832K e820-std off off std 0 0 0
>
> Is there something wrong?
There is something wrong, memtest should run in the loop, and it
should not freeze. If this is a reason of something in your HW or
something in memtest86, I can't tell.
As someone said, you could try http://www.memtest.org/, and see what
it says (I didn't know that it exists).
BR, Jani
--
Jani Averbach
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello,
I downloaded and ran memtest86.
I dont know whether its an error or not. But at top right hand column there
are two sections: Pass and Test.
Now, everytime I run the check, the TEST column goes from 1 to 100% and and
then it again goes back to 2% and it just stops. None of the keys work i.e.
Esc,(C)ancel and the two other keys. I need to reboot my machine.
This is the final report that is being shown after the freeze:
Pentium 4 2400 Mhz
| Pass 1%
L1 Cache 8K 19670 MB/s
| Test 2%
L2 Cache 512K 16782 MB/s
| Test [Address Test, own address, no cache]
Memory 511M 88 MB/s
| Testing 80K 511 M 511M
Chipset /* amazingly its blank but I have a geniune Intel motherboard */
Walltime Cached RsvdMem MemMap Cache ECC
Test Pass Errors ECC Errs
0:0012 511M 832K e820-std off off std 0 0 0
Is there something wrong?
Regards,
Ritesh
----- Original Message -----
From: "Jani Averbach" <ja...@jaa.iki.fi>
To: <us...@subversion.tigris.org>
Sent: Wednesday, November 10, 2004 1:33 PM
Subject: Re: DB corruption of first day of subversion
> On 2004-11-10 13:08+0530, Ritesh Nadhani wrote:
>> On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
>> >Hello,
>> >
>> >We plan to move all our repository from CVS to subversion.
>> >
>> >We are using WinXP with SP2. I was successfuly able to create create a
>> >repository and even was able to import my test project to the
>> >repository.
>> >Our server is running as a standalone svnserve daemon listening on 3690
>> >port. It is on 512Kbps dsl line (bandwith is pretty costly out here).
>> >
>> >One of our developers was also able to download the test project (around
>> >5
>> >MB with documenation and all) to his home machine. He did some
>> >modification
>> >to the code and checked in. All this process were OK.
>> >
>> >Now, suddenly looks like the whole berkely DB is corrupt. Whenever we
>> >are
>> >now trying to create a new project in the repository - we are getting an
>> >error: Berkely DB error while allocating new node ID (getting
>> >'next-key' )
>> >for filesystem f:\svnrepos/db, DB_NOTFOUND: No matching key/data pair
>> >found.
>> >
>> >If I try to checkout a project, we get error:
>> >
>> >Reference to non-existent node 'd.0.1' in filesystem 'F:/svnrepos/db'
>> >
>> >What should I do? Is SVN known to have bugs with Windows XP SP2.
>> >
>> >Regards,
>> >Ritesh
>> >
>> >>>>>>>>
>>
>> F: is just a partition on my hard-disk. I have 4 partition on the box.
>>
>> c:\windows - XP Files
>> d:DATA - General data files
>> e:softwares - all softwares are installed out here instead of \program
>> files\ . Makes it easy to manage.
>> f:\repos - I am using it only for Subversion repository so that backup
>> are
>> easy to do.
>>
>> This drive is not network mapped. I mean you cannot access it from any
>> machine in the networked. The only folder on this machine that can be
>> accessed by other people is the SHareDocs which is present in C;\. Disk
>> management shows it as:
>>
>> Type: Local disk
>> Filesystem: NTFS.
>>
>> Am I missing something?
>
> No, SVN should work on winXP OK. Couple of questions more,
>
> 1) What version of Subversion you are using?
>
> 2) Is the machine otherwise stable? Any flaky memory? Have you tried
> http://www.memtest86.com/ on it? Subversion could push quite a lot
> of stress to the I/O sub system and memory, so it is possible that it
> might trigger random memory errors on unstable HW.
>
> BR, Jani
>
> --
> Jani Averbach
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by "Alex R. Mosteo" <al...@mosteo.com>.
Ritesh Nadhani wrote:
> Hello,
>
> Even if set it up running thru Apache, wont it give the same error. My
> subversion reporsitory got corrupted so does it matter if I use SVN or
> HTTP. Just wanted to know?
>
> Looks like my memory sticks are flaky as memtest86 runs correctly on
> other machines without any problem. It checking one machine as I write
> this and it has reached 75% with it. In my box it only reached 2% and
> just freezed :((
I don't think it should freeze. I've tested bad memory with this program
and it echoes the mismatching patterns but doesn't hang. A heating
problem will cause lock-ups too. Though if the lock-up is consistently
at the very same point of the test then it is either a hardware problem
or some incompatibility with memtest68.
Did you try memtest86+ (http://www.memtest.org)?
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello,
Even if set it up running thru Apache, wont it give the same error. My
subversion reporsitory got corrupted so does it matter if I use SVN or HTTP.
Just wanted to know?
Looks like my memory sticks are flaky as memtest86 runs correctly on other
machines without any problem. It checking one machine as I write this and it
has reached 75% with it. In my box it only reached 2% and just freezed :((
Regards,
Ritesh
----- Original Message -----
From: "Martin Tomes" <li...@tomes.org>
To: "Ritesh Nadhani" <ri...@webyog.com>
Cc: <us...@subversion.tigris.org>
Sent: Wednesday, November 10, 2004 4:33 PM
Subject: Re: DB corruption of first day of subversion
> Ritesh Nadhani wrote:
>> ----- Original Message ----- From: "Jani Averbach" <ja...@jaa.iki.fi>
>> To: <us...@subversion.tigris.org>
>> Sent: Wednesday, November 10, 2004 1:33 PM
>> Subject: Re: DB corruption of first day of subversion
>>> On 2004-11-10 13:08+0530, Ritesh Nadhani wrote:
>>>> On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
>>>> >Hello,
>>>> >
>>>> >We plan to move all our repository from CVS to subversion.
>
> You would do rather better if you set up an Apache based server. I see
> you are using IIS, you could also run Apache on a different port (say 81)
> by setting the Listen configuration paramter to 81 and connect to
> http://your.machine:81/svn/... (whatever url you choose).
>
> I have updated my installation instructions for getting an Apache based
> Subversion server up and running on Windows:
>
> http://www.subversionary.org/subversionary.cgi/ServerSetupWindows
>
> If you do what it says and set Listen to 81 (or whatever you choose) you
> should be ok.
>
> Obviously this isn't the only way to use Subversion but I have found it to
> be the path of least resistance to getting a functioning setup which
> continues to work.
>
> --
> Martin Tomes
> echo 'martin at tomes x org x uk'\
> | sed -e 's/ x /\./g' -e 's/ at /@/'
>
> The Subversion Wiki is at http://www.subversionary.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Martin Tomes <li...@tomes.org>.
Ritesh Nadhani wrote:
> ----- Original Message ----- From: "Jani Averbach" <ja...@jaa.iki.fi>
> To: <us...@subversion.tigris.org>
> Sent: Wednesday, November 10, 2004 1:33 PM
> Subject: Re: DB corruption of first day of subversion
>> On 2004-11-10 13:08+0530, Ritesh Nadhani wrote:
>>> On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
>>> >Hello,
>>> >
>>> >We plan to move all our repository from CVS to subversion.
You would do rather better if you set up an Apache based server. I see you are using IIS, you could
also run Apache on a different port (say 81) by setting the Listen configuration paramter to 81 and
connect to http://your.machine:81/svn/... (whatever url you choose).
I have updated my installation instructions for getting an Apache based Subversion server up and
running on Windows:
http://www.subversionary.org/subversionary.cgi/ServerSetupWindows
If you do what it says and set Listen to 81 (or whatever you choose) you should be ok.
Obviously this isn't the only way to use Subversion but I have found it to be the path of least
resistance to getting a functioning setup which continues to work.
--
Martin Tomes
echo 'martin at tomes x org x uk'\
| sed -e 's/ x /\./g' -e 's/ at /@/'
The Subversion Wiki is at http://www.subversionary.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Ritesh Nadhani <ri...@webyog.com>.
Hello,
----- Original Message -----
From: "Jani Averbach" <ja...@jaa.iki.fi>
To: <us...@subversion.tigris.org>
Sent: Wednesday, November 10, 2004 1:33 PM
Subject: Re: DB corruption of first day of subversion
> On 2004-11-10 13:08+0530, Ritesh Nadhani wrote:
>> On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
>> >Hello,
>> >
>> >We plan to move all our repository from CVS to subversion.
>> >
>> >We are using WinXP with SP2. I was successfuly able to create create a
>> >repository and even was able to import my test project to the
>> >repository.
>> >Our server is running as a standalone svnserve daemon listening on 3690
>> >port. It is on 512Kbps dsl line (bandwith is pretty costly out here).
>> >
>> >One of our developers was also able to download the test project (around
>> >5
>> >MB with documenation and all) to his home machine. He did some
>> >modification
>> >to the code and checked in. All this process were OK.
>> >
>> >Now, suddenly looks like the whole berkely DB is corrupt. Whenever we
>> >are
>> >now trying to create a new project in the repository - we are getting an
>> >error: Berkely DB error while allocating new node ID (getting
>> >'next-key' )
>> >for filesystem f:\svnrepos/db, DB_NOTFOUND: No matching key/data pair
>> >found.
>> >
>> >If I try to checkout a project, we get error:
>> >
>> >Reference to non-existent node 'd.0.1' in filesystem 'F:/svnrepos/db'
>> >
>> >What should I do? Is SVN known to have bugs with Windows XP SP2.
>> >
>> >Regards,
>> >Ritesh
>> >
>> >>>>>>>>
>>
>> F: is just a partition on my hard-disk. I have 4 partition on the box.
>>
>> c:\windows - XP Files
>> d:DATA - General data files
>> e:softwares - all softwares are installed out here instead of \program
>> files\ . Makes it easy to manage.
>> f:\repos - I am using it only for Subversion repository so that backup
>> are
>> easy to do.
>>
>> This drive is not network mapped. I mean you cannot access it from any
>> machine in the networked. The only folder on this machine that can be
>> accessed by other people is the SHareDocs which is present in C;\. Disk
>> management shows it as:
>>
>> Type: Local disk
>> Filesystem: NTFS.
>>
>> Am I missing something?
>
> No, SVN should work on winXP OK. Couple of questions more,
>
> 1) What version of Subversion you are using?
My setup file says: svn-1.1.1-setup-2.exe ( i believe its 1.1.1). I am using
TortoiseSVN 1.1.1 for working with it.
>
> 2) Is the machine otherwise stable? Any flaky memory? Have you tried
> http://www.memtest86.com/ on it? Subversion could push quite a lot
> of stress to the I/O sub system and memory, so it is possible that it
> might trigger random memory errors on unstable HW.
Yes. I never had a crash before. I use it for developing/coding etc. It runs
IIS, MySQL 4.1.7 and some other daemons. Though these are all used for test
cases. I will try memtest86 and report if anything is found here.
But I want to move from CVS to subversion . I just fear that I dont lose all
my data with data corruption.
Regards,
Ritesh
>
> BR, Jani
>
> --
> Jani Averbach
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
> For additional commands, e-mail: users-help@subversion.tigris.org
>
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: DB corruption of first day of subversion
Posted by Jani Averbach <ja...@jaa.iki.fi>.
On 2004-11-10 13:08+0530, Ritesh Nadhani wrote:
> On 2004-11-10 11:45+0530, Ritesh Nadhani wrote:
> >Hello,
> >
> >We plan to move all our repository from CVS to subversion.
> >
> >We are using WinXP with SP2. I was successfuly able to create create a
> >repository and even was able to import my test project to the repository.
> >Our server is running as a standalone svnserve daemon listening on 3690
> >port. It is on 512Kbps dsl line (bandwith is pretty costly out here).
> >
> >One of our developers was also able to download the test project (around 5
> >MB with documenation and all) to his home machine. He did some
> >modification
> >to the code and checked in. All this process were OK.
> >
> >Now, suddenly looks like the whole berkely DB is corrupt. Whenever we are
> >now trying to create a new project in the repository - we are getting an
> >error: Berkely DB error while allocating new node ID (getting 'next-key' )
> >for filesystem f:\svnrepos/db, DB_NOTFOUND: No matching key/data pair
> >found.
> >
> >If I try to checkout a project, we get error:
> >
> >Reference to non-existent node 'd.0.1' in filesystem 'F:/svnrepos/db'
> >
> >What should I do? Is SVN known to have bugs with Windows XP SP2.
> >
> >Regards,
> >Ritesh
> >
> >>>>>>>>
>
> F: is just a partition on my hard-disk. I have 4 partition on the box.
>
> c:\windows - XP Files
> d:DATA - General data files
> e:softwares - all softwares are installed out here instead of \program
> files\ . Makes it easy to manage.
> f:\repos - I am using it only for Subversion repository so that backup are
> easy to do.
>
> This drive is not network mapped. I mean you cannot access it from any
> machine in the networked. The only folder on this machine that can be
> accessed by other people is the SHareDocs which is present in C;\. Disk
> management shows it as:
>
> Type: Local disk
> Filesystem: NTFS.
>
> Am I missing something?
No, SVN should work on winXP OK. Couple of questions more,
1) What version of Subversion you are using?
2) Is the machine otherwise stable? Any flaky memory? Have you tried
http://www.memtest86.com/ on it? Subversion could push quite a lot
of stress to the I/O sub system and memory, so it is possible that it
might trigger random memory errors on unstable HW.
BR, Jani
--
Jani Averbach
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org