TOC when FAIL_FAST not enabled

[Steve Bonds]

Under what circumstances will a ServiceGuard managed node TOC itself when neither NODE_FAIL_FAST_ENABLED nor SERVICE_FAIL_FAST_ENABLED are set?

Pages 75-76 of the Managing ServiceGuard manual (B3936-90045) imply that the host should TOC only if one of the above parameters are set. However, I already discovered that a node timeout will cause the node to TOC itself even when neither of these are set. What else will do it?

[Santosh Nair_1 · Accepted solution]

I believe the node will TOC if cmcld fails or if there is loss of two heartbeat signals.

There is a better explanation in the HP manual, Managin MC/ServiceGuard in the section entitled: Reponse to failure:

http://docs.hp.com/cgi-bin/fsearch/framedisplay?top=/hpux/onlinedocs/B3936-90024/B3936-90024_top.html&con=/hpux/onlinedocs/B3936-90024/00/00/18-con.html&toc=/hpux/onlinedocs/B3936-90024/00/00/18-toc.html&searchterms=node%20timeout&queryid=20011001-113814

-Santosh

[Steve Bonds]

Thanks! I initially read through that part thinking they meant this applied when NODE_FAIL_FAST_ENABLED was set, but indeed it seems to apply globally.

[Stephen Doud]

This topic is dealt with in depth in this ITRC customer-viewable document:
UXSGLVKBAN00000010
" Why Did MC/ServiceGuard Initiate a TOC (crash)?"

There are many reasons that are best answered by the document.
-Stephen