[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: lock scalability

From: Ben Collins-Sussman <sussman_at_collab.net>
Date: 2005-03-14 21:13:51 CET

On Mar 14, 2005, at 1:55 PM, John Peacock wrote:

> Greg Hudson wrote:
>> (c) I'm not personally going to be very sad if such people find that
>> locking large numbers of files consumes large amounts of resources.
>> In
>> my mind, mass file locking isn't a very good way of doing work, so if
>> the system pushes back on people who try it, that's not a terrible
>> thing.
> My reading of what Sussman said:
>> * just like property lists, both sides of the network currently hold
>> unbounded lists of lock-objects in memory.
> suggests that too many locks could potentially crash the server and/or
> client. Has anyone done a "back of the napkin" calculation to see at
> what point the list of lock-objects is likely to exceed the existing
> memory footprint of Subversion? What is sizeof(lock-object)?

cmpilato and I are guessing that on average, one of these will be 500

typedef struct svn_lock_t
   const char *path; /* the path this lock applies to */
   const char *token; /* universally unique URI representing
lock */
   const char *owner; /* the username which owns the lock */
   const char *comment; /* (optional) description of lock */
   svn_boolean_t xml_comment; /* whether comment is packaged in XML
by DAV */
   apr_time_t creation_date; /* when lock was made */
   apr_time_t expiration_date; /* (optional) when lock will expire;
                                    If value is 0, lock will never
expire. */
} svn_lock_t;

To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Mar 14 21:15:54 2005

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.