Since a few weeks our soap webservice went very slow.
I do not know what has changed, maybe the underlaying tomcat version with a linux update.
We have a testbox suite that run real webrequests to the local host, erlier it took 60s for all tests, after migrating to aws it wun in about 45s, and since we moved to a larger machine and used 8GB ram for lucee, we could run all testcases in about 30s.
Now we got complains of the webservice clients that everything went very slow.
My Testcases do actually taking over 200s now.
I tryed several settings and lucee versions.
It was long time running on the old 5.2.x version prior the axis extension.
Actually i can run it also on 3.3.x, i tryed Axcis 24 / 25 / 27 all seems to be the same.
I checked with the fusion reactor support and they said it is lucee internally wastimng the time by mapping my input objects ( only strings., numbers and dates ) into the java objects.
When i observe the single requests, they all taking more then 400ms, and inside is mostly one sql statement which takes 1 - 4 ms… The rest ( 80 - 95% of the time ) lucee is just convertimng my arguments.
The code for the most of the webservice methods is very easy stuff, just arguments vaklidating, and fire one or 2 sql queries on oracle. FR show us that the SQL is allways quick.
What can be the cause of this behaviour ?
I never tryed to change the underlaying tomcat, which i think it’s possiblöy updated by the linux apt update.
Switching to REST is a long term alternative, but we need a solution now, since our data load is slowed down to 15 - 20% of the original performance.
We urgently needing support here.
OS: Linux Ubuntu 18 Linux (5.4.0-1056-aws [amd64])
Java Version: 25.181-b13 (1.8.0_181)
Tomcat Version: 8.5.33
Lucee Version: 220.127.116.11 & 18.104.22.168