I observed this very same issue on a two-nodes cluster with one of the nodes powered off. The culprit in my case was DRBD waiting forever for a connection from the other node. DRBD’s LSB header says that corosync should be started after itself and given the unlimited WFC timeout, this never happened.
My solution was to add: wfc-timeout 30; into the common/startup stanza of DRBD configuration. So far, nothing to do with the package itself. On Tue, 24 Jun 2014 04:24:15 +0300 Tapio Salonsaari <ta...@tietovirta.fi> wrote: > Steps to reproduce: > - Reboot host > - Observe that corosync isn't running > - Start corosync with 'service' or directly via /etc/init.d/ without issues > > I attempted to place some debugging messages to init-script, but it looks > like that > host doesn't even attempt to run init-script, even if symlinks and everything > are > in place: -- Marco Fagiolini X U L - mindWerks / Outside Standing Level ZWEI WEB: http://www.xul.it <http://www.xul.it/> TEL: +49/178/9243283 FAX: +39/02/700443139 IM: skype://oxullo <skype://oxullo>
smime.p7s
Description: S/MIME cryptographic signature
