LAE

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dkhuon@gogoair.com

There are times when, because of lack of data / files, a node would abort.

I would like to trap this condition with a mechanism (such as try/catch) but outside of the failing node, before the mayhem and alter the graph flow to somewhere else (not necessary aborting).

I have the feeling I am not the only person would need this ability.

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: stonysmith

If it is because the file doesn't exist.. then use a Directory List node to get a list of files, then a filter to restrict to just the file you are interested in, then a metacheck node to ensure that at least one file exists.

Metacheck has an option to prevent downstream nodes from running if there are zero input records from the Directory List.

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dkhuon@gogoair.com

Unfortunately the input files are at the other side of the infrastructure wall (firewall), and can only be accessed through FTP.

Normally (in my previous works, using different tools), we would use FTP list to detect and compare file existence and date, etc.. to determine whether there is anything new to be processed. I am using the "prototype" Ftp Get which aborts if no input file :-(

Granted I can wait for the new version of Ftp (in Lavastorm 4.6.1), which "might" have a better flexibility in this regards, but I am wondering if someone has generically dealt with situations whereas we know in advance that a node could "fail" (under "un-preventable" circumstances), and would want to take a preventive action in order to control the graph for a more graceful landing (behavior), such as informing users that we don't input (therefore output) for this period.

If such a feature does not currently exist within Lavastorm, I would like to ADVOCATE for it, knowing that in the future I, for one, would need to use it to tame my graph from aborting ungracefully. I noticed some technique of creating dummy static files to deal with missing data, but it can only help when it is applicable. I guess what I am wishing for is an implementation, at node level, of a try-catch mechanism: maybe a try node that can be paired with a catch node and an end-block node, and using the a kind of clocking signal to control the flow within. Or something to that effect.

Thanks anyway, your input and of others are very welcome.

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: ejones

Just for the record, you can trap a graph failure if you execute the graph using the Execute BRX node. So if you wrap the node in a BRX you can simulate what you are wanting.

Because of the difficulty surrounding setting this up, I'm with you on the need to trap node errors as you describe.

Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.

Originally posted by: dkhuon@gogoair.com

Originally posted by: ejones
					

Just for the record, you can trap a graph failure if you execute the graph using the Execute BRX node.  So if you wrap the node in a BRX you can simulate what you are wanting.


Because of the difficulty surrounding setting this up, I'm with you on the need to trap node errors as you describe.
			

Let's hope Lavastorm will provide some enhancements in this area. Thanks.