[prev in list] [next in list] [prev in thread] [next in thread] 

List:       gluster-users
Subject:    [Gluster-users] Abnormally large number of Open files with Glusterfs
From:       krishna () zresearch ! com (Krishna Srinivas)
Date:       2008-12-18 9:22:41
Message-ID: ad4bc5820812180110i7f27c63em3c3a34663cb7703d () mail ! gmail ! com
[Download RAW message or body]

On Thu, Dec 18, 2008 at 2:29 PM, Chandranshu . <chandranshu at gmail.com> wrote:
> Hi,
>
> We didn't have any particular reason to choose this timeout. We had set it
> to larger values initially and then decreased it in steps. Timeouts started
> to happend at a limit of 1.5s. So, we doubled it to 3s. Ever since then, we
> have not encountered timeouts though the system did become slow at times.
> Are there heuristics available to estimate a good timeout settings?

It is just an educated guess on how much timeout shoule be. It is not
safe to keep it as low as 3 secs. We have seen even keeping it as 20
secs getting timed out when there are very heavy operations (when
stress testing with iozone/dbench)

Krishna



[prev in list] [next in list] [prev in thread] [next in thread]