maybe you could do a while loop with checking "mii-diag -s" which will return exit code 2 if link level heartbeat is not yet established. I would raise an ubuntu bug on that. And, ocfs2 could have a sencond initial heartbeat/retry timeout, since it has not joined the cluster yet, no need to panic early. Gruss Bernd
________________________________ From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alexei_Roudnev Sent: Thursday, December 07, 2006 11:00 PM To: Yuval Baruch Cc: Sunil Mushran; [email protected]; [EMAIL PROTECTED] Subject: Re: [Ocfs2-users] DMesg error on startup ... Ok, it is not about 'ended'. What happen, more likely: - Linux runs network service. It enables interfaces (btw, some interfaces comes by Hot PlugIn daemon after in, asyncronously) - ifconfig reports interface UP. In reality, ethernet is in spanning-tree negotiation mode (so it cant send/receive packets next 30 seconds) - network service exists with 0 return code (because it did everything). So system proceed with next service(s) - it starts o2cb in 2 - 5 seconds after it. - so, next 20 - 30 seconds o2cb cant communicate (because switch did not finish spanning tree negotiation yet - it have 30 or 40 second shold time) and so it fail. - when you run it in the very end, it can work. So, (1) fix switch config OR (2) add 'wait when ping will work' into the network script (or better add one more service which do it) OR (3) add sleep 40 into the o2cb No one oif these methods is not bypass. For 'order', see skeleton file in /etc/init.d . But order is corerct out of the box (at leat, relating to the network - it is not correct if you use iSCSI for example). ----- Original Message ----- From: Yuval Baruch <mailto:[EMAIL PROTECTED]> To: Alexei_Roudnev <mailto:[EMAIL PROTECTED]> Cc: Sunil Mushran <mailto:[EMAIL PROTECTED]> ; [email protected] ; [EMAIL PROTECTED] Sent: Thursday, December 07, 2006 1:05 PM Subject: Re: [Ocfs2-users] DMesg error on startup ... thanks for you offer, but this like the addition of "mount -a" in the rc.local are ways to bypass the problem. Is there a way to prevent a init service with lower preference to start before the init service with higher prefernce ended? like s25ocfs2 will start only after s24o2cb will end ... On 12/7/06, Alexei_Roudnev <[EMAIL PROTECTED]> wrote: Add 'sleep 20' into the beginning of ocfs2 start sequence? Btw, check that your network ports are in fast-spanning-tree mode (don't wait 30 seconds when connected). On the Ethernet switch. ----- Original Message ----- From: Yuval Baruch <mailto:[EMAIL PROTECTED]> To: Sunil Mushran <mailto:[EMAIL PROTECTED]> Cc: [email protected] ; [EMAIL PROTECTED] Sent: Thursday, December 07, 2006 7:43 AM Subject: Re: [Ocfs2-users] DMesg error on startup ... I wasn't exact about the previous mail. The real problem is that o2cb does not complete his startup before the ocfs2 start his. the order is as you mentioned s24o2cb -> s25ocfs2. The problem is as I wrote before: some of te FS are mounted and other doesn't, it depends on how fast the o2cb starts. So is there any way fixing this problem ( other than adding " mount -a" to the rc.local ) ? On 12/7/06, Yuval Baruch <[EMAIL PROTECTED] > wrote: Hi, I am encountring with the same problem, I double checked and the startup order is as u mentioned, i.e. S10network -> S24o2cb, but still some ocfs2 FS tries (and failes) to mount before the network comes up. Other FS which tries to mount after the network comes up succeed, the FS which succeed to mount change from boot to boot ( depends on how long it takes to the network to comes up ). Is there any way to resolve this problem other then adding "mount -a" to the rc.local ? 10x On 11/7/06, Sunil Mushran <[EMAIL PROTECTED] > wrote: The startup order should be... ==> network, o2cb, ocfs2. Sébastien CRAMATTE wrote: > Hi > > When my system boot I've got this error in dmesg : > > mount.ocfs2: Unable to access cluster service Cannot initialize cluster > > ---------- > Mounting local filesystems... > kjournald starting. Commit interval 5 seconds > EXT3 FS on hda2, internal journal > EXT3-fs: mounted filesystem with ordered data mode. > kjournald starting. Commit interval 5 seconds > EXT3 FS on hda3, internal journal > EXT3-fs: mounted filesystem with ordered data mode. > mount.ocfs2: Unable to access cluster service Cannot initialize cluster > /dev/hda2 on /home type ext3 (rw,errors=remount-ro) > /dev/hda3 on /var type ext3 (rw,errors=remount-ro) > Cleaning /tmp /var/run /var/lock. > Setting up networking...done. > Setting up IP spoofing protection: rp_filter. > Configuring network interfaces...Set tx off on eth0 [OK] > Set tx off on eth1 [OK] > done. > -------- > > I've got this line in my /etc/fstab > > /dev/drbd0 /cluster ocfs2 _netdev,datavolume,nointr > 0 0 > > So it's logical that try to mount it ... I've tried to put "noauto" > parameter and efectively I haven't got anymore the error but the FS is > not mounted automatically by ocfs2 rc.d script ... > How can I avoid this error on bootup ? > > Regards > > > > _______________________________________________ > Ocfs2-users mailing list > [email protected] > http://oss.oracle.com/mailman/listinfo/ocfs2-users > _______________________________________________ Ocfs2-users mailing list [email protected] http://oss.oracle.com/mailman/listinfo/ocfs2-users ________________________________ _______________________________________________ Ocfs2-users mailing list [email protected] http://oss.oracle.com/mailman/listinfo/ocfs2-users
_______________________________________________ Ocfs2-users mailing list [email protected] http://oss.oracle.com/mailman/listinfo/ocfs2-users
