Hi, is there a way to make a Splunk transaction wait until it has ended, before starting another transaction.
e.g. if I have (with latest results at the top)
a | end |
b | start |
c | start |
d | end |
e | end |
f | start |
g | start |
h | start |
What I get from Splunk here would be transactions: f->e, g->d and b->a.
But what I want is h->e and c->a, so once it's found "start" it then looks for "end", and then looks for the next "start" after that... etc.
Hi @morganj1,
it's possible if you have a field (e.g. transaction_id) to use as a correlation key to correlate events, if you use startswith and/or endswith options it isn't possible.
Anyway, if you have a field to use as key, see a different approach using the stats command instead transaction because the thansaction command is very slow!
Ciao.
Giuseppe
You could also try to transform your data so that it keeps only the first "start" in a row (by using autoregress/streamstats to get previous value and only leave the "start" if it was preceeded by an "end")
If you are sure that those 'in-between' events are useless, you can even use dedup to filter them out.