On Wed, 2016-11-16 at 10:24 -0800, Gordon Messmer wrote:
On 11/16/2016 09:21 AM, Rich Megginson wrote:
> I suggest you file a ticket at
https://fedorahosted.org/389/newticket
> and attach this and the other information for tracking. This doesn't
> seem like an issue that will be easily resolved . . .
OK. Is there any other data I can gather right now? The problem takes
a few months to manifest, but once it does it gets pretty bad. That
gives us narrow, rare windows to get useful data.
What's your ioblocktimeout set to?
How many connections are idle on the server?
Are you seeing OOM behaviour or memory not being released to the OS?
What specs are your servers ie cpu and memory, is it ecc memory?
What kind of disk are they on? Are there issues in dmesg?
Have you configured system activity reporter (sar), and have out from
the same time of disk io, memory usage, cpu etc?
What's your sysctl setup like?
Have you increased file descriptors for Directory Server?
Have you lowered the TCP close wait timeout?
When I hear of problems like this, I'm always inclined to investigate
the host first, as there is a surprising amount that can affect DS from
the host.
--
Sincerely,
William Brown
Software Engineer
Red Hat, Brisbane