6 messages in com.perforce.perforce-userAdministering Perforce, checkpoints
FromSent OnAttachments
Virg...@perforce.com25 Mar 1998 03:19 
Pete...@auspex.com25 Mar 1998 11:04 
TomB...@ebc.ericsson.se25 Mar 1998 23:55 
Fran...@ti.com26 Mar 1998 02:42 
Pete...@auspex.com26 Mar 1998 09:50 
Rich...@netapp.com26 Mar 1998 11:52 
Subject:Administering Perforce, checkpoints
From:Pete...@auspex.com (Pete@auspex.com)
Date:03/26/1998 09:50:55 AM
List:com.perforce.perforce-user

Francois, et. al.,

In particular, how long does it take to checkpoint the database, given its size and the type of platform you use.

I have currently 8 users. My db.have and db.integ are about 5 MB each, and the whole depot contains about 3500 files, a total of 150 MB.

I spend virtually no time administering Perforce. I do weekly checkpoints at night through a cron job.

(Question to heavy users: do you keep old journals/checkpoints or do you delete them after a while)

I should have answered in more detail at first.

Our trees consist of about 10,000 files each, give or take. This represents a sigle product's evolution over the past 6 years. The following shows what the depot files look like (sizewize) after running our bootstrap script, which checks in 29 trees and creates 22 branches in the process. I would say that this represents a significant amount of work, as much or more than our eventual user community of 150 - 200 developers would go through in a week.

I have a cron job that fires nightly and checkpoints and backups up the database and rcs files to an on-line directory incrementally each night, and does a full backup each weekend. For style points, I eject the tape when I'm done.

All this is done on a 250 MHz Sun Enterprise 450, with 128 MB ram. The entire depot, including rcs files is 1,374,174 KB. The depot directory looks like:

- -rw-r--r-- 1 perforce 138064 Mar 25 11:36 auspex.errors - -rw-r--r-- 1 perforce 553021595 Mar 25 11:34 auspex.journal drwxr-xr-x 18 perforce 512 Mar 25 03:25 ax-test drwxr-xr-x 3 perforce 512 Mar 24 19:29 br - -rwxr-xr-x 1 perforce 16384 Mar 25 04:35 db.change - -rwxr-xr-x 1 perforce 8192 Mar 25 04:23 db.counters - -rwxr-xr-x 1 perforce 8192 Mar 24 19:25 db.depot - -rwxr-xr-x 1 perforce 16384 Mar 25 04:35 db.desc - -rwxr-xr-x 1 perforce 8192 Mar 25 04:09 db.domain - -rwxr-xr-x 1 perforce 8192 Mar 23 23:41 db.fix - -rwxr-xr-x 1 perforce 8192 Mar 23 23:41 db.fixrev - -rwxr-xr-x 1 perforce 2781184 Mar 25 04:35 db.have - -rwxr-xr-x 1 perforce 101433344 Mar 25 03:42 db.integ - -rwxr-xr-x 1 perforce 8192 Mar 23 22:56 db.job - -rwxr-xr-x 1 perforce 8192 Mar 23 22:56 db.jobpend - -rwxr-xr-x 1 perforce 2887680 Mar 25 04:35 db.locks - -rwxr-xr-x 1 perforce 8192 Mar 23 22:56 db.protect - -rwxr-xr-x 1 perforce 64733184 Mar 25 04:35 db.rev - -rwxr-xr-x 1 perforce 36831232 Mar 25 04:35 db.revcx - -rwxr-xr-x 1 perforce 8192 Mar 25 11:34 db.user - -rwxr-xr-x 1 perforce 8192 Mar 25 04:09 db.view - -rwxr-xr-x 1 perforce 5783552 Mar 25 04:35 db.working - -rwxr-xr-x 1 perforce 535 Mar 23 14:48 license

I don't have timing information for this (test) server. For our production server, with only 6.75 MB of data, the entire checkpoint and backup-to- online-directories process (based upon p4backup) takes only 2.5 minutes.

I do nightly checkpoint/backup iterations, because I never want to lose more than a days work.

Peter ===================== Peter DiPrete Consultant, ITE Corp pdiprete at auspex.com Office: (408)566-2432 Cellular: (408)221-1601 Pager (numeric): (888)533-3727, PIN# 4082211601 Pager (text): 4082211601 at paging.cellone-sf.com