On Tue, 2014-06-10 at 17:31 +0000, Moussa Ba (moussaba) wrote: > I have been dealing with a vexing issue where reboots on a CentOS 6.5 where reboots take 30 to 45 minutes. > My configuration: > > * 32 LUNS and 32 network portals. > * 2 ConnectX3 iser > * Observed on 3.12.19, reproduced on target-pending. > > What seems to be happening is that targets/LUNS are released after the > network interfaces are brought down, this leads to what I would assume > to be time outs and long release time for each LUN/target. > > Running /etc/init.d/target stop works as expected. > > Has this been observed on other distributions? > I've not seen this with RoCE + iSER on debian, namely because of: root@ivytown:/# ls -la /etc/rc6.d/K01target lrwxrwxrwx 1 root root 16 May 15 15:25 /etc/rc6.d/K01target -> ../init.d/target root@ivytown:/# ls -la /etc/rc6.d/K06networking lrwxrwxrwx 1 root root 20 May 15 15:06 /etc/rc6.d/K06networking -> ../init.d/networking > How can we change the behavior such that targets/LUNS are released > before network interfaces are brought down. > So reordering the /etc/init.d/networking + /etc/init.d/target in rc6.d would be the easiest solution.. Sagi + Or, have you guys seen this type of strangeness before where shutting down other services atop RoCe takes long amounts of time..? --nab -- To unsubscribe from this list: send the line "unsubscribe target-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
