Re: Odd performance problem (hitching)

From: Don Lewis (truckman_at_FreeBSD.org)
Date: 08/30/05

  • Next message: Noritoshi Demizu: "Re: panic: sackhint rexmit == 0" 
    Date: Mon, 29 Aug 2005 19:50:27 -0700 (PDT)
To: doconnor@gsoft.com.au

    On 30 Aug, Daniel O'Connor wrote:
    > On Tuesday 30 August 2005 03:45, Don Lewis wrote:
    >> Any more info on this problem? I just got my MFC reminder for this
    >> change and I don't want to request permission for the MFC until I know
    >> for sure that it (or 1.636, which is also has a pending MFC) isn't the
    >> cause of the problem.
    >
    > I backed out that change and have observed no change whatsoever.
    >
    >> What else is happening on machine when the symptoms occur? I would only
    >> expect vfs_subr.c 1.636 and 1.642 to affect the behaviour if there was a
    >> lot of file system activity that put pressure on the system to free up
    >> vnodes.
    >
    > It's my workstation/laptop so.. "a variety of things" :)
    >
    > The hitching happens even when the system is idle so I think you can MFC
    > safely.

    I'd be most suspicious of the MNT_VNODE_FOREACH loop in ffs_sync(),
    especially if the symptoms occur every 30 seconds. Running the CPU at
    slow speed (with powerd, etc.) is likely to aggravate the problem. What
    is the value of "sysctl kern.maxvnodes"?

    You might try significantly decreasing MAXVNODES_MAX in
    sys/kern/vfs_subr.c and rebuilding your kernel.

    _______________________________________________
    freebsd-current@freebsd.org mailing list
    http://lists.freebsd.org/mailman/listinfo/freebsd-current
    To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

  • Next message: Noritoshi Demizu: "Re: panic: sackhint rexmit == 0"