Jun
07

9-Current Testing Snapshot 20110607 Now Available‏

The next PC-BSD 9-CURRENT snapshot is now available for download.

This snapshot includes a lot of fixes and changes, notably:

  • Overhauled Life-Preserver backup utility
  • Fixes to AppCafe browser navigation
  • Improvements to system installer pages, such as disk / user setup
  • Improved tray notifier for available updates
  • New “Welcome” program to introduce new users to PC-BSD
  • Wifi tool automatically loaded in tray
  • New beastie splash menu which allows booting from failsafe kernel
  • Improvements for IPv6 support in base / utilities
  • Misc bugfixes reported by users from last snapshot

As usual, please test and report bugs any problems back to the testing mailing list.

Share This Post:
  • Digg
  • Facebook
  • Twitter
  • email
  • LinkedIn
  • Slashdot

Written by dru. Posted in 9.0, testers

Trackback from your site.

Comments (9)

  • George
    June 8, 2011 at 12:10 pm |

    I am testing this on VMWare Fusion 3.1.1, (OSX 10.5, MacBookPro3.1).
    Every time I install some app from AppCafe (FireFox, Chromium) my computer gets hot and noisy after that, fans go crazy, for quite a long period of time. The reason could be linked to somthing around the daemon that manages the PBI hash folder. Something inefficient must be happening, or maybe something loops, I cant tell.
    Of course it could be the vmware machine, but is anyone else having this issue? PCBSD 8.2 runs in the same VM without the problem.

    Basically all I can see is a total of 52% (system) of CPU used (via top), but the most CPU consuming process is “sh” with just units of CPU consumed. Same in Task Manager (all checked in View menu). Am I not seeing all the processess, or is this like during compilation, where new instances of sh appear and disappear so quickly that you cannot catch them in task manager?
    Sorry for the “lama” description, I am new to FreeBSD/PCBSD.

  • June 8, 2011 at 3:00 pm |

    Your description does sound right, most likely it is the PBI daemon doing the merge which is increasing CPU usage. However it is running “niced” and isn’t spawning “sh” processes, so it shouldn’t hinder other desktop usage. I’ll take a look into that though, see if there is anything else I can do to decrease CPU usage at that phase of the installation.

    • George
      June 9, 2011 at 6:20 am |

      Just to explain a big further, the ~52% CPU usage does not matter because of its number (52), but because it stays on that number for an unusually long time. Unusually long with respect to my assumption that the merge task is/should not be anything that involved, what would require the CPU to run high for a long time. Which process should I be looking for?

      • June 9, 2011 at 7:02 am |

        It is the main “sh” process that is using the most CPU time. Sometimes the hash-list is 5-10k of files, and it is just processing through that list, moving and linking files.

        I’ve done some optimization on this though, and it should run a bit quicker for the next release, on my little 1.6ghz atom system a merge which used to take 5 min, took 1.3 min now, so a definite improvement. I’m going to keep tinkering on this though, try to get it as slimmed down as possible for release.

        • George
          June 9, 2011 at 9:35 am |

          Great!

          Question #2: if I understand it right, the hash-sums are calculated at install-time, aren’t they?
          I just assume this, because http://​wiki​.pcbsd​.org/​i​n​d​e​x​.​p​h​p​/​P​B​I​9​_​F​o​r​m​a​t​#​M​a​n​a​g​e​d​_​h​a​r​d​-​l​i​n​k​s​_​f​o​r​_​r​e​s​o​u​r​ces does not explicitly say so.
          How about doing it at PBI-creation time instead (so that PBI would contain the hashes pre-calculated)? Would save all the trouble, if I am not completely wrong about this.

          Question #3: Is it possible to share /usr/pbi (containing the installed apps) between multiple networked machines (NFS mounted), with read-only access?

          • June 9, 2011 at 10:46 am |

            The checksums are generated at creation time, so that isn’t an issue. However I’ve been able to optimize it further today, and now the same 5 minute process is taking about 70 seconds on my slow atom here, probably much faster on any kind of decent hardware.

            As for NFS sharing, sure that would be possible. Just have to make sure the clients are running the same major version of PC-BSD, I.E. 9.0, 9.1, etc.

            If you want to have access to commands like pbi_icon and such on the clients, you would need to nfs mount /var/db/pbi as well, read-only of course :)

  • ben
    June 20, 2011 at 4:39 am |

    waarom niet in het nederlands?????
    zijn wij een te klein landje.
    willen ook wel eens wat andes dan windows

    • June 20, 2011 at 4:57 am |

      Why not what in Dutch?

Leave a comment

*

Please leave these two fields as-is:

Help the Project, Donate Today!