# See the file LICENSE for redistribution information. # # Copyright (c) 2001,2007 Oracle. All rights reserved. # # $Id: rep066.tcl,v 12.10 2007/05/24 20:20:42 alanb Exp $ # # TEST rep066 # TEST Replication and dead log handles. # TEST # TEST Run rep_test on master and a client. # TEST Simulate client crashes (master continues) until log 2. # TEST Open 2nd master env handle and put something in log and flush. # TEST Downgrade master, restart client as master. # TEST Run rep_test on newmaster until log 2. # TEST New master writes log records, newclient processes records # TEST and 2nd newclient env handle calls log_flush. # TEST New master commits, newclient processes and should succeed. # TEST Make sure 2nd handle detects the old log handle and doesn't # TEST write to a stale handle (if it does, the processing of the # TEST commit will fail). # proc rep066 { method { niter 10 } { tnum "066" } args } { source ./include.tcl if { $is_windows9x_test == 1 } { puts "Skipping replication test on Win 9x platform." return } # Run for all access methods. if { $checking_valid_methods } { return "ALL" } set args [convert_args $method $args] set logsets [create_logsets 2] # Run the body of the test with and without recovery. foreach r $test_recopts { foreach l $logsets { set logindex [lsearch -exact $l "in-memory"] if { $r == "-recover" && $logindex != -1 } { puts "Rep$tnum: Skipping\ for in-memory logs with -recover." continue } puts "Rep$tnum ($method $r):\ Replication and dead log handles." puts "Rep$tnum: Master logs are [lindex $l 0]" puts "Rep$tnum: Client logs are [lindex $l 1]" rep066_sub $method $niter $tnum $l $r $args } } } proc rep066_sub { method niter tnum logset recargs largs } { global testdir global rep_verbose set verbargs "" if { $rep_verbose == 1 } { set verbargs " -verbose {rep on} " } env_cleanup $testdir replsetup $testdir/MSGQUEUEDIR set masterdir $testdir/MASTERDIR set clientdir $testdir/CLIENTDIR file mkdir $masterdir file mkdir $clientdir # Log size is small so we quickly create more than one. # The documentation says that the log file must be at least # four times the size of the in-memory log buffer. set pagesize 4096 append largs " -pagesize $pagesize " set log_max [expr $pagesize * 8] set m_logtype [lindex $logset 0] set c_logtype [lindex $logset 1] # In-memory logs require a large log buffer, and cannot # be used with -txn nosync. set m_logargs [adjust_logargs $m_logtype] set c_logargs [adjust_logargs $c_logtype] set m_txnargs [adjust_txnargs $m_logtype] set c_txnargs [adjust_txnargs $c_logtype] # Open a master. # Later we'll open a 2nd handle to this env. repladd 1 set ma_envcmd "berkdb_env_noerr -create $m_txnargs \ $m_logargs -errpfx ENV0 -log_max $log_max $verbargs \ -home $masterdir -rep_transport \[list 1 replsend\]" set env0 [eval $ma_envcmd $recargs -rep_master] set masterenv $env0 # Open a client. repladd 2 set cl_envcmd "berkdb_env_noerr -create $c_txnargs \ $c_logargs -errpfx ENV1 -log_max $log_max $verbargs \ -home $clientdir -rep_transport \[list 2 replsend\]" set env1 [eval $cl_envcmd $recargs -rep_client] set clientenv $env1 # Bring the clients online by processing the startup messages. set envlist "{$env0 1} {$env1 2}" process_msgs $envlist # Run a modified test001 in the master (and update clients). puts "\tRep$tnum.a.0: Running rep_test in replicated env." eval rep_test $method $masterenv NULL $niter 0 0 0 0 $largs process_msgs $envlist set nstart $niter set last_client_log [get_logfile $env1 last] set stop 0 while { $stop == 0 } { puts "\tRep$tnum.b: Run test on master until log file changes." eval rep_test\ $method $masterenv NULL $niter $nstart $nstart 0 0 $largs incr nstart $niter replclear 2 set last_master_log [get_logfile $masterenv last] if { $last_master_log > $last_client_log } { set stop 1 } } # Open a 2nd env handle on the master. # We want to have some operations happen on the normal # handle and then flush them with this handle. puts "\tRep$tnum.c: Open 2nd master env and flush log." set 2ndenv [eval $ma_envcmd -rep_master -errpfx 2NDENV] error_check_good master_env [is_valid_env $2ndenv] TRUE set testfile2 "test$tnum.2.db" set omethod [convert_method $method] set txn [$masterenv txn] error_check_good txn [is_valid_txn $txn $masterenv] TRUE set db [eval {berkdb_open_noerr -env $masterenv -errpfx MASTER \ -txn $txn -create -mode 0644} $largs $omethod $testfile2] error_check_good dbopen [is_valid_db $db] TRUE # Flush on the 2nd handle set lf [stat_field $2ndenv log_stat "Times log flushed to disk"] error_check_good flush [$2ndenv log_flush] 0 set lf2 [stat_field $2ndenv log_stat "Times log flushed to disk"] error_check_bad log_flush $lf $lf2 # The detection of dead log handle is based on a 1-second resolution # timestamp comparison. Now that we've established the threatening # source of the dead handle in $2ndenv, wait a moment to make sure that # the fresh handle that we're about to create gets a later timestamp. tclsleep 1 # Resolve the txn and close the database error_check_good commit [$txn commit] 0 error_check_good close [$db close] 0 # Nuke those messages for client about to become master. replclear 2 puts "\tRep$tnum.d: Swap envs" set masterenv $env1 set clientenv $env0 error_check_good downgrade [$clientenv rep_start -client] 0 error_check_good upgrade [$masterenv rep_start -master] 0 set envlist "{$env0 1} {$env1 2}" process_msgs $envlist # # At this point, env0 should have rolled back across the log file. # We need to do some operations on the master, process them on # the client (but not a commit because that flushes). We want # the message processing client env (env0) to put records in # the log buffer and the 2nd env handle to flush the log. # puts "\tRep$tnum.e: Run test until create new log file." # # Set this to the last log file the old master had. # set last_client_log $last_master_log set last_master_log [get_logfile $masterenv last] set stop 0 while { $stop == 0 } { puts "\tRep$tnum.b: Run test on master until log file changes." eval rep_test\ $method $masterenv NULL $niter $nstart $nstart 0 0 $largs process_msgs $envlist incr nstart $niter set last_master_log [get_logfile $masterenv last] if { $last_master_log == $last_client_log } { set stop 1 } } puts "\tRep$tnum.f: Create some log records." set txn [$masterenv txn] error_check_good txn [is_valid_txn $txn $masterenv] TRUE set db [eval {berkdb_open_noerr -env $masterenv -errpfx MASTER \ -txn $txn -create -mode 0644} $largs $omethod $testfile2] error_check_good dbopen [is_valid_db $db] TRUE process_msgs $envlist # Flush on the 2nd handle puts "\tRep$tnum.g: Flush on 2nd env handle." set lf [stat_field $2ndenv log_stat "Times log flushed to disk"] error_check_good flush [$2ndenv log_flush] 0 set lf2 [stat_field $2ndenv log_stat "Times log flushed to disk"] error_check_bad log_flush2 $lf $lf2 # Resolve the txn and close the database puts "\tRep$tnum.h: Process commit on client env handle." error_check_good commit [$txn commit] 0 error_check_good close [$db close] 0 process_msgs $envlist error_check_good cl2_close [$2ndenv close] 0 error_check_good env0_close [$env0 close] 0 error_check_good env1_close [$env1 close] 0 replclose $testdir/MSGQUEUEDIR return }