Hi All,
I have been running a series of comparison tests from jmeter to K6 and am finding that using the same injection infrastructure am running out of disk space on my injectors 
I do see that for each call I make is resulting in around 12 lines in the output csv file.
http_reqs, http_req_duration, http_req_blocked, http_req_connecting, http_req_tls_handshaking, http_req_sending, http_req_waiting, http_req_receiving, http_req_failed, MyTrendTransaction, checks, checks.
Was hoping to be able to pretty much only capture metrics I have added as trends and included the above details (and response body size) into a single record THEN suppress the rest.
Is this possible?
Unfortunately, this is currently not possible 
It’s high on our priority list, but there are some prerequisite changes we need to make before we get to it… Follow Add explicit tracking and ignoring of metrics and sub-metrics · Issue #1321 · grafana/k6 · GitHub for updates, and maybe share your specific use case there as a data point?
@BobRuub have you tried using compressing the csv data through adding .gz to the end this will likely elevate some of the disk requirements.
Other than that, I can only recommend that you play with fifo pipes and filter things this way:
mkfifo pipe
cat < pipe | grep 'http_req_duration' >new.csv & // change filtering to what you want, you can also compress through gzip here as well
k6 run -o csv=pipe script.js
rm pipe // some cleanup
k6 will write EOF so the cat/grep filtering process will also stop with k6.
Hope this helps.
Thanks for the response.
I’ll have a crack at the piping solution as it looks like the best bet in the short term.
It’s high on our priority list, but there are some prerequisite changes we need to make before we get to it… Follow