On Fri, Dec 12, 2014 at 3:42 AM, Haster <[email protected]> wrote:
>
> Hi guys!
>
> We have a problem with my Couchbase claster (2.5.1) on Linux server
>
> When we start to fill bucket with data one of node is failed with such
> error: "Port server memcached exit with status 137"
>
> I've googled this error and found some discussion there guys said that
> problem could be in OOM Killer.
>
> So, how can I fix this problem.
> How can I guaranty that OOM Killer won't kill my couchbase process?
>
> In documentation I didn't find any recommendations of how much of total
> memory I can use for Couchbase.
> Is there any formula?
>

Hi.

Yes it is likely OOM. Because exit status 137 means that process was killed
by SIGKILL. And if you haven't sent this signal to memcached, then it could
be only kernel. In later case it's almost certainly due to OOM. You can
verify that by looking at dmesg on that box. If that's OOM, the log will
have detailed explanation from the kernel why it had to pick memcached.

Regarding memory sizing I'm sure there was some documentation. And in
principle if you stick to defaults (60% of ram), you should be fine in most
cases. Default leaves usually sufficient space for page cache and memory
fragmentation overhead, although in some artificial tests it was shown to
be not enough.

Can you share more details of your setup? Like ram size, quotas etc ?

>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Couchbase" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to