Re: [jgroups-users] general questions about FD_ALL

--===============8768000690005642578==
Content-Type: multipart/alternative; boundary="000000000000aea56805abfc8005"

--000000000000aea56805abfc8005
Content-Type: text/plain; charset="UTF-8"

Hi,

Thanks for this. My eventual followup question:

We have some customers who want to have failover (view change in jgroups
terms, plus the extra checks we do after a node fails) times with timeouts
on the order of a couple seconds. I guess that makes sense in some places,
like where you know the network is hard-wired or using containers where
everything is on one machine anyway. Our default failure time is ~1 minute,
but people can try whatever they think is appropriate.

Using FD_ALL, if someone really wants a 5 second timeout, should we just
change the interval to match? Others wanted me to ask if there's anything
magic about 8 seconds. :)

You recently answered a question of mine about FD_SOCK that was very
helpful. Am thinking of setting things up so, if someone wants a really
small timeout, then we turn on FD_SOCK in that case. Anyone who wants a
timeout <8 seconds clearly isn't worried about temporary network hiccups.

Thanks again,
Bobby


On Tue, May 5, 2020 at 2:26 AM Questions/problems related to using JGroups <
[email protected]> wrote:

>
>
> On 01.05.20 19:05, Questions/problems related to using JGroups wrote:
> > Hi,
> >
> > We use a TCP-based stack, and use FD_ALL for failure detection. This is
> > with jgroups 4.1.8. The only property we set in FD_ALL is "timeout,"
> > which is configurable by the end user.
>
> No, there's also interval and timeout_check_interval. See [1] for details
>
> > It looks from the code like the
> > heartbeat message each node sends out is 8 seconds apart -- is that
> > correct?
>
> Yes, that's defined by 'interval'
>
> > If someone sets the "timeout" value to something low, e.g.
> > lower than 8 seconds, how does jgroups handle it?
>
> There's a warning: [2]
>
> > Am getting questions about the timing of checks, how long it might take
> > to notice a node is down (e.g. timeout+8 seconds?), etc, and wanted to
> > check on these.
>
> If you have FD_SOCK in the stack, then the default time to exclude a
> crashed member is ~1.5s (VERIFY_SUSPECT).
>
> If not, only FD_ALL, with timeout=40s and interval=8s (defaults), then
> ~41.5s (40 + 1.5).
>
> [1] http://www.jgroups.org/manual4/index.html#FD_ALL
>
> [2]
>
> https://github.com/belaban/JGroups/blob/master/src/org/jgroups/protocols/FD_ALL.java#L59
>
> >
> > Thanks,
> > Bobby
> >
> >
> >
> > _______________________________________________
> > javagroups-users mailing list
> > [email protected]
> > https://lists.sourceforge.net/lists/listinfo/javagroups-users
> >
>
> --
> Bela Ban | http://www.jgroups.org
>
>
>
> _______________________________________________
> javagroups-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/javagroups-users
>

--000000000000aea56805abfc8005
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi,<div><br></div><div>Thanks for this. My eventual follow=
up question:</div><div><br></div><div>We have some customers who want to ha=
ve failover (view change in jgroups terms, plus the extra checks we do afte=
r a node fails) times with timeouts on the order of a couple seconds. I gue=
ss that makes sense in some places, like where you know the network is hard=
-wired or using containers where everything is on one machine anyway. Our d=
efault failure time is ~1 minute, but people can try whatever they think is=
 appropriate.</div><div><br></div><div>Using FD_ALL, if someone really want=
s a 5 second=C2=A0timeout, should we just change the interval to match? Oth=
ers wanted me to ask if there&#39;s anything magic=C2=A0about 8 seconds. :)=
</div><div><br></div><div>You recently answered a question of mine about FD=
_SOCK that was very helpful. Am thinking of setting things up so, if someon=
e wants a really small timeout, then we turn on FD_SOCK in that case. Anyon=
e who wants a timeout &lt;8 seconds clearly isn&#39;t worried about tempora=
ry network hiccups.</div><div><br></div><div>Thanks again,</div><div>Bobby<=
/div><div><br></div></div><br><div class=3D"gmail_quote"><div dir=3D"ltr" c=
lass=3D"gmail_attr">On Tue, May 5, 2020 at 2:26 AM Questions/problems relat=
ed to using JGroups &lt;<a href=3D"mailto:[email protected]=
e.net">[email protected]</a>&gt; wrote:<br></div><bloc=
kquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left:=
1px solid rgb(204,204,204);padding-left:1ex"><br>
<br>
On 01.05.20 19:05, Questions/problems related to using JGroups wrote:<br>
&gt; Hi,<br>
&gt; <br>
&gt; We use a TCP-based stack, and use FD_ALL for failure detection. This i=
s <br>
&gt; with jgroups 4.1.8. The only property we set in FD_ALL is &quot;timeou=
t,&quot; <br>
&gt; which is configurable by the end user.<br>
<br>
No, there&#39;s also interval and timeout_check_interval. See [1] for detai=
ls<br>
<br>
&gt; It looks from the code like the <br>
&gt; heartbeat message each node sends out is 8 seconds apart -- is that <b=
r>
&gt; correct? <br>
<br>
Yes, that&#39;s defined by &#39;interval&#39;<br>
<br>
&gt; If someone sets the &quot;timeout&quot; value to something low, e.g. <=
br>
&gt; lower than 8 seconds, how does jgroups handle it?<br>
<br>
There&#39;s a warning: [2]<br>
<br>
&gt; Am getting=C2=A0questions about the timing of checks, how long it migh=
t take <br>
&gt; to notice a node is down (e.g. timeout+8 seconds?), etc, and wanted to=
 <br>
&gt; check on these.<br>
<br>
If you have FD_SOCK in the stack, then the default time to exclude a <br>
crashed member is ~1.5s (VERIFY_SUSPECT).<br>
<br>
If not, only FD_ALL, with timeout=3D40s and interval=3D8s (defaults), then =
<br>
~41.5s (40 + 1.5).<br>
<br>
[1] <a href=3D"http://www.jgroups.org/manual4/index.html#FD_ALL" rel=3D"nor=
eferrer" target=3D"_blank">http://www.jgroups.org/manual4/index.html#FD_ALL=
</a><br>
<br>
[2] <br>
<a href=3D"https://github.com/belaban/JGroups/blob/master/src/org/jgroups/p=
rotocols/FD_ALL.java#L59" rel=3D"noreferrer" target=3D"_blank">https://gith=
ub.com/belaban/JGroups/blob/master/src/org/jgroups/protocols/FD_ALL.java#L5=
9</a><br>
<br>
&gt; <br>
&gt; Thanks,<br>
&gt; Bobby<br>
&gt; <br>
&gt; <br>
&gt; <br>
&gt; _______________________________________________<br>
&gt; javagroups-users mailing list<br>
&gt; <a href=3D"mailto:[email protected]" target=3D"_b=
lank">[email protected]</a><br>
&gt; <a href=3D"https://lists.sourceforge.net/lists/listinfo/javagroups-use=
rs" rel=3D"noreferrer" target=3D"_blank">https://lists.sourceforge.net/list=
s/listinfo/javagroups-users</a><br>
&gt; <br>
<br>
-- <br>
Bela Ban | <a href=3D"http://www.jgroups.org" rel=3D"noreferrer" target=3D"=
_blank">http://www.jgroups.org</a><br>
<br>
<br>
<br>
_______________________________________________<br>
javagroups-users mailing list<br>
<a href=3D"mailto:[email protected]" target=3D"_blank"=
>[email protected]</a><br>
<a href=3D"https://lists.sourceforge.net/lists/listinfo/javagroups-users" r=
el=3D"noreferrer" target=3D"_blank">https://lists.sourceforge.net/lists/lis=
tinfo/javagroups-users</a><br>
</blockquote></div>

--000000000000aea56805abfc8005--


--===============8768000690005642578==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline


--===============8768000690005642578==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
javagroups-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/javagroups-users

--===============8768000690005642578==--