Segment failure ( postmaster.pid file is missing )

[rakesh7185]

Hi ,

i have installed greenplum Single Node Edition on suse linux with 4 segments.
due to abrupt shutdown of the linux server we have encountered disk errors.
then recovered the disk using 'fsck' command and then the linux server started fine.but from then on, i was unable to start the greenplum database with the segments.

'postmaster.pid' files are missing in the segments.

i am able to start only the gp master.

tried using the pg_ctl restart ,pg_ctl stop , pg_ctl start providing the master data directory but had no luck so far.

it says- pg_ctl: could not start server

how do i bring up the database with all the segments workin fine??

when i stop the database with the 'gpstop' command we encounter the following message

-------------------------------------------------
Failed Segment Stop Information
-------------------------------------------------

DBID:2 FAILED host:'linux-tuz4' datadir:'/seg-1/gpsne0' with reason:'Shutdown failed'
-----------------------------------------------------
Segments stopped successfully = 3
Segments with errors during stop = 1
-----------------------------------------------------
Successfully shutdown 3 of 4 segment instances
-----------------------------------------------------
Failed to shutdown 1 of 4 segment instances
A total of 1 errors were encountered


to recover failed segment -1 i issued the command gprecoverseg then i got the following message.

-gprecoverseg failed. (Reason='could not connect to server: Connection refused
Is the server running on host "127.0.0.1" and accepting
TCP/IP connections on port 5432?

please help me out in this regard.




Regards,
Rakesh

[carlo.ascani@2ndquadrant.it]

Hi Rakesh,

keep in mind that gprecoverseg requires the presence of a mirror segment to recover from.

If you have not mirroring enabled, simply correct the problem on the segment and restart the server to bring it back online, with gpstop -r

Waiting for feedback on this.

Cheers,

[rakesh7185]

ya mirroring is not enabled in our server.now,how should i bring up the segment -1 back online because...the postmaster.pid file is missing,and without postmaster.pid file the segment ...i cannot start the green plum database along with the segments.


can i create a new postmaster.pid file...and bring up the database...??


please suggest me how to recover from this issue.


Regards,
Rakesh

[ebegoli]

It is possible that your database went down in a unstable state, so your transaction log (binary file in pg_xlog) might be pointing to
a process that does not exist any more. Also, your pg_control file might be corrupted.

You might need to reset pg_xlog:

pg_resetxlog -f $MASTER_DATA_DIRECTORY

WARNING: this step will reset your transaction log and it might reset your pg_control file.
Please make sure you make backups of your database and proceed at your own risk.

Hope this helps,
Edmon