What is the easiest/recommended way to find out when a checkpoint is done when r

shamec over 9 years ago

What is the easiest/recommended way to find out when a checkpoint is done when run for all engines? In my scenario we have nightly cron job that runs the checkpoint-suite script and we're trying to run some other scripts(cleanup/backup) after the checkpoint is complete and to do that we need to get a good idea of how much the checkpointing is taking so we can schedule the other scripts accordingly. To do that we need to monitor the checkpointing for a time and get an average duration so I need to know what our options are for doing that? Can it be done programmatically v.s. manually?

Discussion posts and replies are publicly visible

0 scottb471
Appian Employee
over 9 years ago

I'd recommend wrapping the checkpoint in a shell script that would give the date / time before the checkpoint, call the checkpoint, then give the date / time thereafter and pipe each of these commands to the same log file. The script would look something like this...

date >> ~/appian/logs/checkpoint.log
$HOME/.bash_profile; cd $HOME/appian/server/_scripts; ./checkpoint-suite.sh >> ~/appian/logs/checkpoint.log
date >> ~/appian/logs/checkpoint.log
echo "" >> ~/appian/logs/checkpoint.log
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 shamec over 9 years ago

Scott, this approach assumes that the checkpoint-suite script won't return until the actual checkpointing is done - is that a valid assumption?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 scottb471
Appian Employee
over 9 years ago

Sorry. You are correct that the checkpoint command does not wait until completion to return to the shell. You'd have to set up a file monitor using something like inotify in Linux to check when the file was created after it was started.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 shamec over 9 years ago

Thanks Scott - I will also try Appian Support and see what they recommend.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Eduardo Fuentes
Appian Employee
over 9 years ago

Shame,

You can get an average by opening your db_* logs and searching for this keyword

tailsave

the amount of milliseconds is logged in that line. This is an example of checkpoint taking 369 ms

WARN .a.pf.te "" "369" ".a.db.tailsave"

Usually the engines that'll take longer to checkpoint are: EXEC, DESIGN and ANALYTICS depending on which one is larger on memory/disk and the usage.
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel