[p4] Managing diskspace usage of Perforce Proxy

Frank Compagner frank.compagner at guerrilla-games.com
Thu May 31 04:48:52 PDT 2007


Actually, the proxy does stop working. I'd almost started to doubt
the user reports, but I did just try it out, and when the proxy gets
full, the user can no longer do sync's or anything else that would add
files to the proxy cache. p4 either gives several warnings ("Proxy could
not update its cache.  File is ...") before giving up, or it returns
an actual error (eg: "write: depot/depot/.../File.bin,d\t3860t874849.tmp:
There is not enough space on the disk."). In the first case some files
still might get synced, but the sync is aborted pretty soon. As far as
the user is concerned, Perforce has stopped working.

----------------------------------------------------------------
Frank Compagner                                  Guerrilla Games

DL> The proxy does not stop working, it just gets slow because every file
DL> then comes from the server.  It sorta looks like its not working
DL> because it spits out error msgs when the cache becomes full.

DL> dave

DL> On 5/25/07, paul goffin <paul.goffin at dsl.pipex.com> wrote:
>> Quoting Frank Compagner <frank.compagner at guerrilla-games.com>:
>>
>> > But the main point is that a proxy server should be almost completely
>> > maintenace free. In the current situation, if the admin forgets to
>> > check the proxy regularly (happened once or twice), the proxy fills up,
>> > and perforce stops working for the users on that proxy. This is just
>> > stupid, and I want something more low maintenance.
>>
>> I agree.  Have you reported this to Perforce as a Bug?
>>
>> I can see implementing cache-ageing features on the proxy would be a big
>> overhead for it, but if the cache is full the proxy should fall back to working
>> as a relay.  It shouldn't stop working completely!
>> _______________________________________________
>> perforce-user mailing list  -  perforce-user at perforce.com
>> http://maillist.perforce.com/mailman/listinfo/perforce-user
>>


More information about the perforce-user mailing list