[p4] What to do when the depot reaches a particular MB size

Robert Cowham robert at vaccaperna.co.uk
Fri Oct 7 03:35:59 PDT 2005


Shiv Sikand presented some ideas around this that he has implemented for
ICManage at the Perforce User Conference. Look up his paper (very similar to
what Jeff has mentioned below). 

> -----Original Message-----
> From: perforce-user-bounces at perforce.com 
> [mailto:perforce-user-bounces at perforce.com] On Behalf Of Jeff Grills
> Sent: 06 October 2005 20:47
> To: 'Russell C. Jackson'; venussoftop at hotpop.com; 
> perforce-user at perforce.com
> Subject: RE: [p4] What to do when the depot reaches a 
> particular MB size
> 
> 
> Large depots don't necessarily have to keep all the data for 
> your depot at the same level of reliability/availability.  
> For example, if you are running a perforce server on a UNIX 
> machine with an expensive RAID subsystem, you could migrate 
> some of depot files to a slower disk and use symbolic links 
> to get from the RAID to the slower disk.  Throw in a standard 
> 500gb disk or two and you should have all the room you need 
> for files you're keeping around only for archival.  If you 
> didn't want to pollute your nice server machine with some 
> unreliable storage devices, you could even point those 
> symbolic links to a file system mounted on another machine.
> 
> As a word of caution, I'd suggest doing this for directories 
> within perforce that you do not want to change, and enforce 
> the read-only nature of those files through your protection 
> table and also your operating system by mounting the file 
> system in read-only mode.  You wouldn't even need to 
> regularly backup those depot files because you know they 
> can't be modified.
> You certainly want at least one backup you know to be good - 
> perhaps another drive (or two) that is an exact image of the 
> active drive, so that you can swap it into the machine 
> quickly if the active drive dies.
> 
> Before you head down this path, though, you'd want to compare 
> the cost of setting all that infrastructure up versus simply 
> obtaining more storage in your high availability server.
> 
> j




More information about the perforce-user mailing list