>> Do you shutdown machine gracefully? Thanks for bring up this point. AmazonEC2Client.stopInstances command is utilized to shutdown machines.
After reviewing the stopInstances docs, the Amazon shutdowns seem to be uncoordinated. "Before stopping an instance, make sure it is in a state from which it can be restarted." Question: Is there a remote command that can be given to gracefully shutdown SEDNA, or do I need to create a backdoor command? >> Send this script, please. The following is the script: #!/bin/bash ############################################################ # Sedna Database Service # Copyright (c) 2008 Charles Foster, http://www.cfoster.net/ # # Start, stop, restart or query the status # of the Sedna XML Database. # # This service manages the both the Sedna Governor Server # as well as loading/unloading child database instances. # # You are free to modify and redistribute this script, # providing you leave the copyright message in tact. ############################################################ # Home of The Sedna Installation SEDNA_HOME=/home/ubuntu/sedna # The user account which will run Sedna USER=ubuntu # Database instances to start/stop, separated by spaces DBLIST="qbdb" ############################################################ # Don't edit below here, unless you know what you are doing. ############################################################ SPIDX="pidof se_gov" DESC="Sedna XML Database Server" IDESC="Sedna DB" SEGOV="/etc/init.d/sedna" . /lib/lsb/init-functions do_start() { if [ `$SPIDX` ]; then log_daemon_msg "$DESC already running." else log_daemon_msg "Starting $DESC" sudo -u $USER nohup $SEDNA_HOME/bin/se_gov > /dev/null;\ if [ `$SPIDX` ]; then log_end_msg 0 else log_end_msg 1 fi fi } do_start_dbs() { for n in $DBLIST;\ do\ log_daemon_msg "Starting $IDESC \"$n\"" if [ ! -d $SEDNA_HOME/data/"$n"_files ]; then log_end_msg 1 else if sudo -u $USER nohup $SEDNA_HOME/bin/se_sm $n > /dev/null; then log_end_msg 0 else log_end_msg 1 fi fi done } do_stop_dbs() { if [ ! `$SPIDX` ]; then return fi for n in $DBLIST;\ do\ log_daemon_msg "Stopping $IDESC \"$n\"" if [ ! -d $SEDNA_HOME/data/"$n"_files ]; then log_end_msg 1 else if sudo -u $USER nohup $SEDNA_HOME/bin/se_smsd $n > /dev/null; then log_end_msg 0 else log_end_msg 1 fi fi done } get_status() { if [ `$SPIDX` ]; then $SEDNA_HOME/bin/se_rc else echo "$DESC not running." fi } do_stop() { log_daemon_msg "Stopping $DESC" if [ ! `$SPIDX` ]; then log_daemon_msg "$DESC was not running in the first place." log_end_msg 1 else sudo -u $USER nohup $SEDNA_HOME/bin/se_stop > /dev/null if [ ! `$SPIDX` ]; then log_end_msg 0 else log_end_msg 1 fi fi } case "$1" in start) do_start do_start_dbs ;; status) get_status ;; stop) do_stop_dbs do_stop ;; restart) do_stop_dbs do_stop do_start do_start_dbs ;; *) echo "Usage: $SEGOV {start|stop|status|restart}" >&2 exit 1 ;; esac exit $? From: Ivan Shcheklein [mailto:shchekl...@gmail.com] Sent: Sunday, November 13, 2011 4:13 AM To: Malcolm Davis Cc: sedna-discussion@lists.sourceforge.net Subject: Re: [Sedna-discussion] SEDNA WaitForProcess failed Hi Malcolm, >From time to time, after an Amazon instances has been shutdown, SEDNA fails to start properly. Do you shutdown machine gracefully? I use a modified init.d script provided by Charles Foster to start and stop SEDNA. Send this script, please. Command line Starting database recovery or hot-backup restoration... Database recovery in progress... SEDNA Message: FATAL ERROR in process [720] /home/ubuntu/sedna/bin/se_sm qbDb System error. This error means system malfunction. Cannot create data directory for soft fault logs Cannot create soft fault log ...... Send exact commands you run. On behalf of which user. Send also result of: ls -lh SEDNA_HOME ls -lh SEDNA_HOME/data ls -lh SEDNA_HOME/cfg ERROR 11/11/2011 15:30:13 (RCV qbDb pid=594) [vmm.cpp:vmm_remap:104]: (SE1035) Memory mapping error. INFO 11/11/2011 15:30:13 (SM qbDb pid=548) [bm_core.cpp:bm_log_out_io_statistics:521]: IO block reads:34, writes:0 LOG 11/11/2011 15:30:13 (RCV qbDb pid=594) [tr.cpp:TRmain:514]: Session is closed SYS 11/11/2011 15:30:14 (GOV pid=536) [uprocess.c:uWaitForChildProcess:495]: waitpid (code = 10): No child processes FATAL 11/11/2011 15:30:14 (GOV pid=536) [gov_table.cpp:wait_erase_session:133]: Error, WaitForProcess failed Is this log snippet right after OS start or after your attempt to start database via command line? Is there something I can do to recover from the failure? First of all, we should understand what's going on. Output and log you sent contain a log of strange things. BTW, probably you can give us copy of your machine to reproduce this? Is there a standard start and stop practice that should be utilized? Yep. se_stop or se_stop -hard to terminate immediately all running transactions must be called to shutdown Sedna server gracefully. Is this problem resolved in the newest SEDNA RC? Again it's hard to say since I don't understand problem right now. Though, RC contains a lot of fixes including in recovery process. So, give it a try. Ivan Shcheklein, Sedna Team ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Sedna-discussion mailing list Sedna-discussion@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sedna-discussion
