libdb/test/tcl/repmgr023.tcl
2011-09-13 13:44:24 -04:00

183 lines
5.5 KiB
Tcl

# See the file LICENSE for redistribution information.
#
# Copyright (c) 2009, 2011 Oracle and/or its affiliates. All rights reserved.
#
# TEST repmgr023
# TEST Test of JOIN_FAILURE event for repmgr applications.
# TEST
# TEST Run for btree only because access method shouldn't matter.
proc repmgr023 { { niter 50 } { tnum 023 } args } {
source ./include.tcl
if { $is_freebsd_test == 1 } {
puts "Skipping replication manager test on FreeBSD platform."
return
}
# QNX does not support fork() in a multi-threaded environment.
if { $is_qnx_test } {
puts "Skipping repmgr$tnum on QNX."
return
}
set method "btree"
set args [convert_args $method $args]
puts "Repmgr$tnum ($method): Test of JOIN_FAILURE event."
repmgr023_sub $method $niter $tnum $args
}
proc repmgr023_sub { method niter tnum largs } {
global testdir
global util_path
global databases_in_memory
global repfiles_in_memory
global rep_verbose
global verbose_type
set verbargs ""
if { $rep_verbose == 1 } {
set verbargs " -verbose {$verbose_type on} "
}
set repmemargs ""
if { $repfiles_in_memory } {
set repmemargs "-rep_inmem_files "
}
env_cleanup $testdir
file mkdir [set dira $testdir/SITE_A]
file mkdir [set dirb $testdir/SITE_B]
file mkdir [set dirc $testdir/SITE_C]
foreach { porta portb portc } [available_ports 3] {}
# 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_buf [expr $pagesize * 2]
set log_max [expr $log_buf * 4]
set cmda "berkdb_env_noerr -create -txn nosync \
$verbargs $repmemargs -rep -thread \
-log_buffer $log_buf -log_max $log_max -errpfx SITE_A \
-home $dira"
set enva [eval $cmda]
$enva repmgr -timeout {connection_retry 5000000} \
-local [list localhost $porta] -start master
set cmdb "berkdb_env_noerr -create -txn nosync \
$verbargs $repmemargs -rep -thread \
-log_buffer $log_buf -log_max $log_max -errpfx SITE_B \
-home $dirb"
set envb [eval $cmdb]
$envb repmgr -timeout {connection_retry 5000000} \
-local [list localhost $portb] -start client \
-remote [list localhost $porta]
puts "\tRepmgr$tnum.a: wait for client B to sync with master."
await_startup_done $envb
set cmdc "berkdb_env_noerr -create -txn nosync \
$verbargs $repmemargs -rep -thread \
-log_buffer $log_buf -log_max $log_max -errpfx SITE_C \
-home $dirc"
set envc [eval $cmdc]
$envc repmgr -timeout {connection_retry 5000000} \
-local [list localhost $portc] -start client \
-remote [list localhost $porta]
puts "\tRepmgr$tnum.b: wait for client C to sync with master."
await_startup_done $envc
# Clobber replication's 30-second anti-archive timer, which will have
# been started by client sync-up internal init, so that we can do a
# log_archive in a moment.
#
$enva test force noarchive_timeout
# Run rep_test in the master.
puts "\tRepmgr$tnum.c: Running rep_test in replicated env."
set start 0
eval rep_test $method $enva NULL $niter $start 0 0 $largs
incr start $niter
puts "\tRepmgr$tnum.d: Close client."
set last_client_log [get_logfile $envc last]
$envc close
set stop 0
while { $stop == 0 } {
# Run rep_test in the master.
puts "\tRepmgr$tnum.e: Running rep_test in replicated env."
eval rep_test $method $enva NULL $niter $start 0 0 $largs
incr start $niter
puts "\tRepmgr$tnum.f: Run db_archive on master."
$enva log_flush
$enva test force noarchive_timeout
set res [eval exec $util_path/db_archive -d -h $dira]
set first_master_log [get_logfile $enva first]
if { $first_master_log > $last_client_log } {
set stop 1
}
}
puts "\tRepmgr$tnum.g: Restart client."
set envc [eval $cmdc -recover -event]
$envc rep_config {autoinit off}
$envc repmgr -timeout {connection_retry 5000000} \
-local [list localhost $portc] -start client \
-remote [list localhost $porta]
# Since we've turned off auto-init, but are too far behind to sync, we
# expect a join_failure event.
#
await_condition {[expr [stat_field $envc rep_stat "Startup complete"] \
|| [is_event_present $envc join_failure]]}
error_check_good failed [is_event_present $envc join_failure] 1
# Do a few more transactions at the master, and see that the client is
# still OK (i.e., simply that it's still accessible, that we can read
# data there), although of course it can't receive the new data.
#
puts "\tRepmgr$tnum.h: Put more new transactions, which won't\
get to client C"
eval rep_test $method $enva NULL $niter $start 0 0 $largs
incr start $niter
# Env a (the master) will match env b, but env c will not match.
rep_verify $dira $enva $dirb $envb 0 1 1
rep_verify $dira $enva $dirc $envc 0 0 1
if {$databases_in_memory} {
set dbname { "" "test.db" }
} else {
set dbname "test.db"
}
set dbp [eval berkdb open \
-env $envc [convert_method $method] $largs $dbname]
set dbc [$dbp cursor]
$dbc get -first
$dbc get -next
$dbc get -next
set result [$dbc get -next]
error_check_good got_data [llength $result] 1
error_check_good got_data [llength [lindex $result 0]] 2
$dbc close
$dbp close
# Shut down the master, so as to force client B to take over. Since we
# didn't do any log archiving at B, client C should now be able to sync
# up again.
#
puts "\tRepmgr$tnum.i: Shut down master, client C should sync up."
$enva close
await_startup_done $envc 40
$envc close
$envb close
set test_be_quiet ""
}