Intermittent network freezes - Solaris

This is a discussion on Intermittent network freezes - Solaris ; I have a Sun Fire x2200M2 with 11 zones. I currently have three interfaces configured. bge0 - Two public IPs and 11 private IPs (for the zones) bge1 - One public IP only used for access to ilom interface nge0 ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Intermittent network freezes

  1. Intermittent network freezes

    I have a Sun Fire x2200M2 with 11 zones. I currently have three
    interfaces configured.

    bge0 - Two public IPs and 11 private IPs (for the zones)
    bge1 - One public IP only used for access to ilom interface
    nge0 - One private IP that is just a cross-over cable to another Sun
    Fire x2100 M2 server on a different private ip range than bge0
    interfaces.

    I am using ipf and ipnat to map public IP ports to the private zones/
    ports. Example. Accessing public IP1 port 53 maps to zone 1 port 53
    (Master DNS), public port IP2 port 53 maps to zone 2 port 53
    (Secondary DNS) and so on through other services, ie: MySQL, Apache,
    etc...

    The odd problem I am having is the the network completely stops
    responding randomly for anywhere between 30 seconds to 2 minutes. The
    server is running, I can access it via ILOM (This interface ALWAYS
    works), touch all the zones, however, I can not ping out or come into
    the public IPs on bge0. Then suddenly it starts responding again.

    The last time this happened, I was able to login via console (ILOM)
    and force a core dump, however, I do not have the knowledge to debug.
    I have the minimum support that I purchased with the servers for
    access to SunSolve patches (I think like $360 for three years), does
    anyone have any idea on what might be the problem or what to maybe
    look for in the core dump? If this is a hardware problem, I would like
    to get the server repaired, if it's software, I need to get it fixed.

    This server runs a production webserver, so the intermittent network
    hangs is causing major problems.


  2. Re: Intermittent network freezes

    In alt.solaris.x86 Xao wrote:
    > I have a Sun Fire x2200M2 with 11 zones. I currently have three
    > interfaces configured.
    >
    > bge0 - Two public IPs and 11 private IPs (for the zones)
    > bge1 - One public IP only used for access to ilom interface
    > nge0 - One private IP that is just a cross-over cable to another Sun
    > Fire x2100 M2 server on a different private ip range than bge0
    > interfaces.
    >
    > I am using ipf and ipnat to map public IP ports to the private zones/
    > ports. Example. Accessing public IP1 port 53 maps to zone 1 port 53
    > (Master DNS), public port IP2 port 53 maps to zone 2 port 53
    > (Secondary DNS) and so on through other services, ie: MySQL, Apache,
    > etc...
    >
    > The odd problem I am having is the the network completely stops
    > responding randomly for anywhere between 30 seconds to 2 minutes. The
    > server is running, I can access it via ILOM (This interface ALWAYS
    > works), touch all the zones, however, I can not ping out or come into
    > the public IPs on bge0. Then suddenly it starts responding again.
    >


    I have an X2200, and have seen similar symptoms. Are you able to talk to
    it via the crossover cable to the other X2200?

    Mine is configured in a single zone, and with a crossover cable to a T1000.
    If I leave it alone for a day and then attempt to talk to it there will be
    a delay similar to what you describe. I can reach it via the ILOM, but via
    either its public IP on bge0, or the private one from the T1000 it hangs
    for maybe a minute or so. When either interface begins to talk, then the
    other also locks up.

    I have found that since I started using ntp to keep the time accurate based
    upon a remote server, this other problem seems to have gone away. It seems
    to me that once the ethernet interfaces have all been asleep for a while it
    takes them some effort to wake up again. I have not managed to spot any
    particular activity which is causing it to wait, and the problem seems to
    exist for all IP traffic. Ntp sends out messages on a regular basis, and
    this appears to be enough to keep the IP stuff alive.


    --
    Dr Tristram J. Scott
    Energy Consultant

  3. Re: Intermittent network freezes

    Hmmm, I'll have to check the private interface, yes it's a cross-over
    between the two servers on the nge0 interface.

    I have to say, you may be on to something with the constant network
    traffic. I upped our monitoring from 5 minutes to 30 seconds this
    weekend so I could catch it with a page, and dump the core on the live
    system to open a ticket with Sun, but of course, since I'm monitoring
    so frequently, it's been stable.

    This seems like a very troubling issue if it's not an oddity.
    Especially since this was happening on my 06/07 Build and my 11/07
    build. :/

    Thanks for the input!

    On Jul 29, 10:10 am, tristram.sc...@ntlworld.com (Tristram Scott)
    wrote:
    > I have an X2200, and have seen similar symptoms. Are you able to talk to
    > it via the crossover cable to the other X2200?
    >
    > Mine is configured in a single zone, and with a crossover cable to a T1000.
    > If I leave it alone for a day and then attempt to talk to it there will be
    > a delay similar to what you describe. I can reach it via the ILOM, but via
    > either its public IP on bge0, or the private one from the T1000 it hangs
    > for maybe a minute or so. When either interface begins to talk, then the
    > other also locks up.
    >
    > I have found that since I started using ntp to keep the time accurate based
    > upon a remote server, this other problem seems to have gone away. It seems
    > to me that once the ethernet interfaces have all been asleep for a while it
    > takes them some effort to wake up again. I have not managed to spot any
    > particular activity which is causing it to wait, and the problem seems to
    > exist for all IP traffic. Ntp sends out messages on a regular basis, and
    > this appears to be enough to keep the IP stuff alive.
    >
    > --
    > Dr Tristram J. Scott
    > Energy Consultant




+ Reply to Thread