Re: Datbase replication questions.

On Mon, 18 Nov 1996 01:37:56 +0000, David Williams <djw@smooth1.demon.co.uk> wrote: > > > I have been experimenting with database replication using Online > 7.12. I have the following DR parameters set: > > DRAUTO 2 (REVERSE_TYPE) > DRINTERVAL -1 (Synchronous updates) > DRLOSTFOUND ./dr.lostfound > DRTIMEOUT 30> > Primary and secondary have indentical onconfig's except for > server name/alias and SERVERNUM's. All paths are relative (I have > scripts to switch betweent the two enviroments). Each server has a > temporary dbspace (Flag TEMP set to "Y" on it) as mentioned in the > manuals. > > and have the following questions:- > > 1. How do you handle the situation where you want to reboot both > machines overnight. E.g. because other applications on the machine > required it to be rebooted each night. > > 2. What happens if both primary and secondary machine go down cleanly > due to a power failure. > > I have tried to simulate this by > > - Checkpointing on primary and waiting until checkpoint complete > appears in both servers online.logs. > - issueing an onmode -ky to both servers (secondary first) within > a few seconds of each other. > > When restarting both (again within a few seconds of each other, > primary first) I have seen several times of behaviour:- > > a) Everything comes up ok. > > b) Secondary gets the following flags on it's logical logs > (this is from memory but you get the idea) > ----L- > --CL-- > --C--- > > i.e. 2 logical logs with L set (Last checkpoint record) > and 2 logical logs with C set (current logical log) > > onstat - indicates it is in fast recovery but it does not > recover (>5 minutes) > > c) Secondary reachs Fast Recovery > > Primary hangs with onstat - showing initialisation > BLocked: CKPT REQD. > > Primary once did start OK but took > 5 minutes. > Other times it does hang for >10 minutes and does not > recover. > > > Can anyone help? This is on a test system at work so I can try out > suggestions (at the moment I have been using oninit -ky to shut down > hung servers and oninit -i to rebuild them!). > > >-- >David Williams You may want to talk with Informix support.... I know that we saw the problem with multiple logs indicated as current checkpoint log. I think this bug and many others are fixed in some of the 7.13XXX versions. Richard richj@iquest.com