You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Karl Fogel <kf...@red-bean.com> on 2008/08/05 21:37:04 UTC
Re: lock on whole repository
Mark K <ma...@mitsein.net> writes:
> when using svnadmin upgrade, I see a message about svnadmin locking the
> repository. From what I can tell in the code, all that happens is that
> svnadmin upgrade implies --wait. Is there actually some sort of
> locking involved?
I'm not sure, but maybe it just means you can't commit while the upgrade
is under way? What happens if you try operations against a repository
while it's being upgraded?
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: lock on whole repository
Posted by Karl Fogel <kf...@red-bean.com>.
Mark K <ma...@mitsein.net> writes:
> The main reason that I was looking for an answer to this is that I
> wrote my own verison of fsfs-reshard since fsfs-reshard.py is marked as
> not fit for production. I wrote one in perl that doesn't attempt to
> handle as many cases as the original fsfs-reshard.py does but I wanted
> to add some sort of locking mechanism to it.
>
> At any rate, I did my upgrade last week and just ran svnadmin
> upgrade /svn/repos && fsfs-reshard.pl /svn/repos on all of my
> repositories and crossed my fingers :). Nothing went wrong, but it
> would be nice to improve the script with some locking.
Or better yet, an 'svnadmin lock' command, that would make the
repository readable but not writeable until 'svnadmin unlock' is
issued. (If someone wants a repository to be neither writeable nor
readable, they can just shut off access at the RA layer -- turn off
svnserve, httpd, whatever.)
I'll ask why fsfs-reshard.py is considered unready for production use,
on dev@.
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: fsfs-reshard.py not ready for prime time?
Posted by Karl Fogel <kf...@red-bean.com>.
"David Glasser" <gl...@davidglasser.net> writes:
> On Sun, Aug 10, 2008 at 8:35 AM, Mark Phippard <ma...@gmail.com> wrote:
>> On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
>>> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
>>> this block appears at the end:
>>>
>>> if __name__ == '__main__':
>>> raise Exception, \
>>> """This script is unfinished and not ready to be used on live data.
>>> Trust us."""
>>> main()
>>>
>>> The exception-raise was added in r24552 by Malcolm:
>>>
>>> * tools/server-side/fsfs-reshard.py
>>> Prevent use, for now, and add a license block.
>>>
>>> Suggested by: cmpilato
>>>
>>> Since then there have been a few commits to fsfs-reshard.py, but the
>>> exception is still in place. Do we know exactly what it is that is
>>> making fsfs-reshard.py be not ready for production use? The reason I
>>> ask is that people are starting to write their own reshard scripts (see
>>> the mail below from users@)...
>>
>> From what I remember, I thought the only issue was that when you
>> wanted to use the script to go back to the 1.4 format it took a really
>> long time.
>
> ISTR it was just never particularly tested or reviewed by more than
> one or two people.
Hrm. The warning message is a lot stronger than that. Should we change
it to
"This script is not widely tested. Please make a backup of your
repository before running it. Run with '--really' to avoid this
warning and make the script actually run."
Or something like that?
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: fsfs-reshard.py not ready for prime time?
Posted by Karl Fogel <kf...@red-bean.com>.
"David Glasser" <gl...@davidglasser.net> writes:
> On Sun, Aug 10, 2008 at 8:35 AM, Mark Phippard <ma...@gmail.com> wrote:
>> On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
>>> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
>>> this block appears at the end:
>>>
>>> if __name__ == '__main__':
>>> raise Exception, \
>>> """This script is unfinished and not ready to be used on live data.
>>> Trust us."""
>>> main()
>>>
>>> The exception-raise was added in r24552 by Malcolm:
>>>
>>> * tools/server-side/fsfs-reshard.py
>>> Prevent use, for now, and add a license block.
>>>
>>> Suggested by: cmpilato
>>>
>>> Since then there have been a few commits to fsfs-reshard.py, but the
>>> exception is still in place. Do we know exactly what it is that is
>>> making fsfs-reshard.py be not ready for production use? The reason I
>>> ask is that people are starting to write their own reshard scripts (see
>>> the mail below from users@)...
>>
>> From what I remember, I thought the only issue was that when you
>> wanted to use the script to go back to the 1.4 format it took a really
>> long time.
>
> ISTR it was just never particularly tested or reviewed by more than
> one or two people.
Hrm. The warning message is a lot stronger than that. Should we change
it to
"This script is not widely tested. Please make a backup of your
repository before running it. Run with '--really' to avoid this
warning and make the script actually run."
Or something like that?
-Karl
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: fsfs-reshard.py not ready for prime time?
Posted by David Glasser <gl...@davidglasser.net>.
On Sun, Aug 10, 2008 at 8:35 AM, Mark Phippard <ma...@gmail.com> wrote:
> On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
>> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
>> this block appears at the end:
>>
>> if __name__ == '__main__':
>> raise Exception, \
>> """This script is unfinished and not ready to be used on live data.
>> Trust us."""
>> main()
>>
>> The exception-raise was added in r24552 by Malcolm:
>>
>> * tools/server-side/fsfs-reshard.py
>> Prevent use, for now, and add a license block.
>>
>> Suggested by: cmpilato
>>
>> Since then there have been a few commits to fsfs-reshard.py, but the
>> exception is still in place. Do we know exactly what it is that is
>> making fsfs-reshard.py be not ready for production use? The reason I
>> ask is that people are starting to write their own reshard scripts (see
>> the mail below from users@)...
>
> From what I remember, I thought the only issue was that when you
> wanted to use the script to go back to the 1.4 format it took a really
> long time.
ISTR it was just never particularly tested or reviewed by more than
one or two people.
--dave
--
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Re: fsfs-reshard.py not ready for prime time?
Posted by David Glasser <gl...@davidglasser.net>.
On Sun, Aug 10, 2008 at 8:35 AM, Mark Phippard <ma...@gmail.com> wrote:
> On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
>> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
>> this block appears at the end:
>>
>> if __name__ == '__main__':
>> raise Exception, \
>> """This script is unfinished and not ready to be used on live data.
>> Trust us."""
>> main()
>>
>> The exception-raise was added in r24552 by Malcolm:
>>
>> * tools/server-side/fsfs-reshard.py
>> Prevent use, for now, and add a license block.
>>
>> Suggested by: cmpilato
>>
>> Since then there have been a few commits to fsfs-reshard.py, but the
>> exception is still in place. Do we know exactly what it is that is
>> making fsfs-reshard.py be not ready for production use? The reason I
>> ask is that people are starting to write their own reshard scripts (see
>> the mail below from users@)...
>
> From what I remember, I thought the only issue was that when you
> wanted to use the script to go back to the 1.4 format it took a really
> long time.
ISTR it was just never particularly tested or reviewed by more than
one or two people.
--dave
--
David Glasser | glasser@davidglasser.net | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: fsfs-reshard.py not ready for prime time?
Posted by Mark Phippard <ma...@gmail.com>.
On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
> this block appears at the end:
>
> if __name__ == '__main__':
> raise Exception, \
> """This script is unfinished and not ready to be used on live data.
> Trust us."""
> main()
>
> The exception-raise was added in r24552 by Malcolm:
>
> * tools/server-side/fsfs-reshard.py
> Prevent use, for now, and add a license block.
>
> Suggested by: cmpilato
>
> Since then there have been a few commits to fsfs-reshard.py, but the
> exception is still in place. Do we know exactly what it is that is
> making fsfs-reshard.py be not ready for production use? The reason I
> ask is that people are starting to write their own reshard scripts (see
> the mail below from users@)...
Re: fsfs-reshard.py not ready for prime time?
Posted by Mark Phippard <ma...@gmail.com>.
On Sun, Aug 10, 2008 at 11:27 AM, Karl Fogel <kf...@red-bean.com> wrote:
> In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
> this block appears at the end:
>
> if __name__ == '__main__':
> raise Exception, \
> """This script is unfinished and not ready to be used on live data.
> Trust us."""
> main()
>
> The exception-raise was added in r24552 by Malcolm:
>
> * tools/server-side/fsfs-reshard.py
> Prevent use, for now, and add a license block.
>
> Suggested by: cmpilato
>
> Since then there have been a few commits to fsfs-reshard.py, but the
> exception is still in place. Do we know exactly what it is that is
> making fsfs-reshard.py be not ready for production use? The reason I
> ask is that people are starting to write their own reshard scripts (see
> the mail below from users@)...
fsfs-reshard.py not ready for prime time?
Posted by Karl Fogel <kf...@red-bean.com>.
In http://svn.collab.net/repos/svn/trunk/tools/server-side/fsfs-reshard.py,
this block appears at the end:
if __name__ == '__main__':
raise Exception, \
"""This script is unfinished and not ready to be used on live data.
Trust us."""
main()
The exception-raise was added in r24552 by Malcolm:
* tools/server-side/fsfs-reshard.py
Prevent use, for now, and add a license block.
Suggested by: cmpilato
Since then there have been a few commits to fsfs-reshard.py, but the
exception is still in place. Do we know exactly what it is that is
making fsfs-reshard.py be not ready for production use? The reason I
ask is that people are starting to write their own reshard scripts (see
the mail below from users@)...
-Karl
> On Tue, 05 Aug 2008 17:37:04 -0400
> Karl Fogel <kf...@red-bean.com> wrote:
>
>> Mark K <ma...@mitsein.net> writes:
>> > when using svnadmin upgrade, I see a message about svnadmin locking
>> > the repository. From what I can tell in the code, all that happens
>> > is that svnadmin upgrade implies --wait. Is there actually some
>> > sort of locking involved?
>>
>> I'm not sure, but maybe it just means you can't commit while the
>> upgrade is under way? What happens if you try operations against a
>> repository while it's being upgraded?
>
> I haven't tried it because an upgrade takes such a short amount of
> time :). Digging more into the code, I think there is some sort of
> fcntl lock through APR on the write-lock file that acts as a marker
> that a transaction is underway.
>
> The main reason that I was looking for an answer to this is that I
> wrote my own verison of fsfs-reshard since fsfs-reshard.py is marked as
> not fit for production. I wrote one in perl that doesn't attempt to
> handle as many cases as the original fsfs-reshard.py does but I wanted
> to add some sort of locking mechanism to it.
>
> At any rate, I did my upgrade last week and just ran svnadmin
> upgrade /svn/repos && fsfs-reshard.pl /svn/repos on all of my
> repositories and crossed my fingers :). Nothing went wrong, but it
> would be nice to improve the script with some locking.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Re: lock on whole repository
Posted by Mark K <ma...@mitsein.net>.
On Tue, 05 Aug 2008 17:37:04 -0400
Karl Fogel <kf...@red-bean.com> wrote:
> Mark K <ma...@mitsein.net> writes:
> > when using svnadmin upgrade, I see a message about svnadmin locking
> > the repository. From what I can tell in the code, all that happens
> > is that svnadmin upgrade implies --wait. Is there actually some
> > sort of locking involved?
>
> I'm not sure, but maybe it just means you can't commit while the
> upgrade is under way? What happens if you try operations against a
> repository while it's being upgraded?
>
I haven't tried it because an upgrade takes such a short amount of
time :). Digging more into the code, I think there is some sort of
fcntl lock through APR on the write-lock file that acts as a marker
that a transaction is underway.
The main reason that I was looking for an answer to this is that I
wrote my own verison of fsfs-reshard since fsfs-reshard.py is marked as
not fit for production. I wrote one in perl that doesn't attempt to
handle as many cases as the original fsfs-reshard.py does but I wanted
to add some sort of locking mechanism to it.
At any rate, I did my upgrade last week and just ran svnadmin
upgrade /svn/repos && fsfs-reshard.pl /svn/repos on all of my
repositories and crossed my fingers :). Nothing went wrong, but it
would be nice to improve the script with some locking.
--
Mark K
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org