Skip to main content

Search

Welcome to Tibero GTS!

DB Down and Boot Process in Case of Abnormal DB Shutdown

Document Type | Technical Information

Category | Backup/Recovery

Document Number | TBATI029

 

Overview

This document provides guidance on the shutdown and startup procedures including logs when the DB is abnormally shut down for a specific reason.
 
 

Method

1. Terminate All DB Connection Sessions

Disconnect all DB connection sessions due to WAS down, etc. (jdown, kill -9 [jeus pid])
 

2. Check Estimated Session Rollback Time

SQL>  select sess_id, start_time, used_blk, EXP_RB_TIME from  v$transaction;

SESS_ID START_TIME                         USED_BLK EXP_RB_TIME
---------- -------------------------------- ---------- -----------
        33 2016-04-22 19:35:32                   142690          970

1 row selected.

 

3. Force DB Shutdown

If the estimated rollback time is too long and you cannot wait for completion, perform a checkpoint and then forcibly shut down the DB.
-- Perform checkpoint

SQL> ALTER SYSTEM CHECKPOINT;

 

4. tbdown abort

 

5. Perform Recovery

Recovery is performed during the next tbboot.
[cache recovery in progress]
  [ACF][0] 01 tbsvr_mo:166 [0] acquires Instance Recovery lock(MOUNT mode)
  [ACF][0] 01 tbsvr_mo:230 [0] holding Instance Recovery locks(RECOVERY mode)
  [CLC][0] 01 tc_recov:1761 [0] TC RECOVERY...PREPARE
  โ€ฆ
  [CLC][0] 01 tc_recov:1249 THREAD#0 logfile scanned. flushed=[582.4194386/0000.02c5dd33].
  [CLC][0] 01 tc_recov:1271 THREAD#0, Recovery start tsn progress.(0000.02c02bdc => 0000.02c075d6)
  [CLC][0] 01 tc_recov:1400 THREAD#0 needs REPLAY, from 0000.02c075d6, (5786040 LOG_BLKs)
  [CLC][0] 01 tc_cf_df:2012 DATAFILES OPEN 0000.02c5dd34
  [CLC][0] 01 tc_recov:1990 [0] TC RECOVERY...START. incar no 0
  โ€ฆ
  [CLC][0] 01 tc_rset.:1382 RSET inits claiming bufs 632142
  [CLC][0] 01 tc_rset.:1451 no more free bh for recovery
  [CLC][0] 01 tc_rset.:1485 complete_ir=0 claiming bufs (reco: 204512/632142, cur: 427630, skip: 0)
[CLC][0] 01 tc_recov:2065 [0] can perform Partial Instance Recovery
[cache recovery 1st pass Done.]
  [CLC][0] 01 tc_recov:2073 [0] IR 1st Pass Done
โ€ฆ
[cache recovery 2nd pass Done.]
  [CLC][0] 01 tc_recov:2018 [0] IR 2nd Pass Done
โ€ฆ
[cache recovery complete]
  [CLC][0] 01 tc_recov:2090 [0] TC RECOVERY...DONE
...
[Tx recovery in progress]
  [CT-][0] 01 tx_recov:1001 TX RECOVERY...PREPARE
  [CT-][0] 01 tx_recov:1094 TX RECOVERY...START (dead 34)
 โ€ฆ
[DB is open and Tx recovery continues on specific sessions.]
  [FRM][0] 01 tbsvr_ma:1496 Server is running at NORMAL mode (bootmode : NORMAL)
โ€ฆ
[Tx recovery complete]
  [CT-][0] 01 tx_recov:696 TX RECOVERY...DONE

Note

If there was a transaction in progress when the session ended, sess_recovering will be performed.
If the tbdown immediate command was used to shut down, the DB will wait until sess_recovering completes before shutting down, and you can confirm that sess_recovering is in progress through messages appearing every 30 seconds in the tracelog.
tbsvr_er:063 THROW. ec=ERROR_SERVER_IN_SHUTDOWN(-12039) [ Server instance is shutting down.   ] (csr_id:4294967295) [tbsvr_wthr.c:2145:wthr_get_new_cli_con]

[Output every 30 seconds]
tbsvr_sh:229 shutdown: waiting for releasing SESS lock (tid=38, thr_status=THR_STAT_WAITING, sess_status=SESS_RECOVERING)
tbsvr_sh:229 shutdown: waiting for releasing SESS lock (tid=38, thr_status=THR_STAT_WAITING, sess_status=SESS_RECOVERING)
โ€ฆ

Caution

Cache recovery cannot be skipped without performing it.
Also, if the instance is stopped and restarted during cache recovery, the process will start over from the beginning.

 

 

 

Related to