|Return to convert_dcs1_to_dcs1x.txt CVS log||Up to [Development] / JSOC / doc|
File: [Development] / JSOC / doc / convert_dcs1_to_dcs1x.txt
Revision: 1.3, Mon Sep 9 19:26:38 2013 UTC (9 years, 6 months ago) by prodtest
CVS Tags: Ver_LATEST, Ver_DRMSLATEST, Ver_9-5, Ver_9-41, Ver_9-4, Ver_9-3, Ver_9-2, Ver_9-1, Ver_9-0, Ver_8-8, Ver_8-7, Ver_8-6, Ver_8-5, Ver_8-4, Ver_8-3, Ver_8-2, Ver_8-12, Ver_8-11, Ver_8-10, Ver_8-1, NetDRMS_Ver_LATEST, NetDRMS_Ver_9-5, NetDRMS_Ver_9-41, NetDRMS_Ver_9-4, NetDRMS_Ver_9-3, NetDRMS_Ver_9-2, NetDRMS_Ver_9-1, NetDRMS_Ver_9-0, NetDRMS_Ver_8-8, NetDRMS_Ver_8-7, NetDRMS_Ver_8-6, NetDRMS_Ver_8-5, NetDRMS_Ver_8-4, NetDRMS_Ver_8-3, NetDRMS_Ver_8-2, NetDRMS_Ver_8-12, NetDRMS_Ver_8-11, NetDRMS_Ver_8-10, NetDRMS_Ver_8-1, HEAD
Changes since 1.2: +3 -1 lines
*** empty log message ***
8/01/2013 dcs1x:/home/prodtest/cvs/JSOC/doc/convert_dcs1_to_dcs1x.txt PRELIMINARY NOTES ------------------- We must keep the name dcs1x. At the end dcs1 will be retired. Alert DDS that we are beginning the transition process. There may be a time when they cannot transfer data. (See if we can mitigate this) Any missed data will automatically be re-requested once the dcs1x is operational. Stop all processing on dcs1: (First figure out what to do about the warmstandby on dcs2?) *Hit Exit on the socdc gui *sum_stop_dc after do backup below: *pg_ctl -D /var/lib/pgsql/dcs1_data stop [-m fast] Backup the dcs1 db. Remember there is no drms on dcs1. Make sure all the current rows in dcs1x db have been purged. We could us this: Begin the backup process on dcs1: > psql -p 5431 -c "select pg_start_backup('2013-07-19')" Create a tar archive of the system data directory: > tar -czvf dcs1_backup.tar.gz dcs1_data/ Stop the backup process on dcs1: > psql -p 5431 -c "select pg_stop_backup()" or since there are only 9 sums tables to transfer, it might be better just to copy like so: jsocdc=# copy sum_file to '/tmp/sum_file.copy'; and on dcs1x after scp the file over: hmidb_sums=# copy sum_file from '/tmp/sum_file.copy'; The files to copy from dcs1 are: Schema | Name | Type | Owner --------+------------------+----------+---------- public | sum_ds_index_seq | sequence | postgres public | sum_file | table | postgres public | sum_group | table | postgres public | sum_main | table | postgres public | sum_open | table | postgres public | sum_partn_alloc | table | postgres public | sum_partn_avail | table | postgres public | sum_seq | sequence | postgres public | sum_tape | table | postgres (9 rows) The drms tables are in place on dcs1x. The defined series are: Name Archive Retention UnitSize -------------------- ------- --------- -------- hmi.lev0_isp_0021 0 60 1 hmi.lev0_seq_0011 0 60 1 hmi.tlm 1 5 1 hmi.lev0a -1 60 12 TBD: Notes on moving the network connection and IP addr. And how we stop and start the dataflow. Copy the dcs1:/SUM1 drwxrwsr-x 61149 root SOI 1609728 Jul 19 10:43 SUM1/ to dcs1x:/SUM1 drwxrwsr-x 51996 sumsadmin sums 1175552 Jun 27 01:11 SUM1/ Populate the T50 with the original dcs1 tapes. NFS /data on dcs1x to cl1n001 To start dcs1x, see: /home/prodtest/cvs/JSOC/doc/new_dcsx_run.txt Enter the crontab -e with the jobs from dcs1. NOTE: This doc s/b developed into a checklist to be followed. NOTE: The dcs2x has both a hmidb[_sums] and a aiadb[_sums], so it can temporarily take on either role if dcs0x or dcs1x is out of service. The concept is that data would be reprocessed when the main machine is back, and tape archiving would proceed from where it left off. We should look closer at this, but an attempt has been made to simplify this host backup from what we had previously (conflicting db names and port numbers and /SUMs names, etc. have been eliminated).