Was the transaction log dir not removed while the iodevices dir was? It seems like the system got in an invalid state where the transaction logs don't agree with the index state. It kind of looks like the iodevices got swapped around or something? I see on one boot the indexes exist in /local_data/dblab/pkgs/..../data/red , and then on the next they exist in ...data/blue .
On Sun, Feb 21, 2021 at 2:16 AM Tin Vu <[email protected]> wrote: > > Hi Ian, > > Please find the NC log in the attachment. > > Actually I was able to set it up using Ansible, which is a more clean way > than using NCService. But it would be still great to figure out what happened > with NCService option. > > Thanks, > > Tin > > On Sun, Feb 21, 2021 at 1:36 AM Ian Maxon <[email protected]> wrote: >> >> Can you post the NC log? That should show why it failed to start. >> >> On Sat, Feb 20, 2021 at 5:04 PM Tin Vu <[email protected]> wrote: >> > >> > Hi all, >> > >> > I'm trying to install AsterixDB 0.9.6 on my cluster. I'm following the >> > tutorial at: >> > https://ci.apache.org/projects/asterixdb/ncservice.html#Multi_machine >> > >> > Currently, I'm able to start NCService on each node. However, when I start >> > the cluster controller using the following command: >> > ``` >> > bin/asterixcc -config-file cc.conf > cc.log 2>&1 & >> > ``` >> > I always got the following error: >> > ``` >> > 16:57:05.115 [Executor-12:ClusterController] ERROR >> > org.apache.asterix.app.replication.NcLifecycleCoordinator - Node >> > asterix_nc1 failed to complete startup >> > org.apache.hyracks.api.exceptions.HyracksDataException: attempt to create >> > metadata index Dataverse. Index should already exist >> > ``` >> > >> > Have any of you got into this problem before? If so, I would appreciate it >> > if you can give me a hint to solve it. I attached the config file for >> > reference. >> > >> > Thanks, >> > >> > Tin
