Home Forums OS X Server and Client Discussion File Serving Anyone here seeing the AFP load issues many Leopard Server admins are seeing?

Viewing 4 posts - 1 through 4 (of 4 total)
  • Author
    Posts
  • #375578
    mcnaugha
    Participant

    There’s a massive discussion on it over at Apple Discussions. Many saying they’ve got AppleCare cases logged. I’ve got a customer who has one logged too. May only affect high impact servers, such as those in high school environments with hundreds of Mac and PC clients. AFP swaps the CPU allowing nothing else to operate normally. This results in all sorts of problems at the client end from login failures to slow saves. It essentially brings the entire school down.

    It almost looked like the recent Security Update was going to fix this, as it referred to a race condition, but sadly it hasn’t helped.

    #376151
    smirq
    Participant

    Defnintely. We’ve seen it in our school environment in 10.4 and 10.5. It was fixed eventually in 10.4, but we are still waiting with 10.5…

    #376461
    dL_EVO
    Participant

    I just noticed this problem yesterday with three of my leopard servers.

    I’m suspecting it’s an action done by one of the users logged into their AFP network home folder that causes the high CPU usage by the AppleFileService process.

    I noticed last night that all of my users logged off and my AppleFileService process was still using about 180% – 200% CPU. Whatever is happening is getting AFP stuck in some kind of loop it appears.

    I have opened a case with Applecare Enterprise.

    I tried the following actions last night and my server seems fine this morning. I don’t know if it’s actually fixed, maybe someone else can try it and let us know..

    Step 1: Turn Off AFP in Server Admin.

    Step 2: Delete or move the following files:

    /etc/afp.conf
    /etc/afpovertcp.cfg

    /Library/Preferences/com.apple.AppleFileServer.plist

    Restart the server. Turn AFP back on.

    #376584
    warrens
    Participant

    We are seeing the same issue, it’s hit all of our home servers, though the older G5 cluster node is by far the most troublesome. Has the above operation proved to be a lasting solution?

Viewing 4 posts - 1 through 4 (of 4 total)
  • You must be logged in to reply to this topic.

Comments are closed