Upgraded from 43.0.3 -> 44.0.0. Pages take much longer to generate

[lazysteel93]

I am running a Prometheus alert-manager instance, and post-upgrade the time to generate each page is significantly slower than before.

Sample from app logs is below. No other dependency changed in between. (Both built this morning, minutes apart).

44.0.0:

alerta-746cf85bcd-ldwzx alerta-backend [pid: 8|app: 0|req: 52/431] 10.76.26.8 () {66 vars in 2663 bytes} [Thu Jan  2 10:32:26 2025] GET /cgi-bin/environments?status=open&status=ack => generated 273 bytes in 3964 msecs (HTTP/1.1 200) 6 headers in 226 bytes (1 switches on core 0)

43.0.3:

alerta-85d77fc9f4-hvtk4 alerta-backend [pid: 11|app: 0|req: 56/155] 10.76.26.8 () {66 vars in 2663 bytes} [Thu Jan  2 10:39:38 2025] GET /cgi-bin/environments?status=open&status=ack => generated 273 bytes in 61 msecs (HTTP/1.1 200) 6 headers in 226 bytes (1 switches on core 0)

Relevant requirements.txt:

[Flask-Compress](https://pypi.org/project/Flask-Compress/)==1.17
[Flask-Cors](https://pypi.org/project/Flask-Cors/)==5.0.0
[Flask](https://pypi.org/project/Flask/)==3.1.0
[MarkupSafe](https://pypi.org/project/MarkupSafe/)==3.0.2
[PyJWT](https://pypi.org/project/PyJWT/)==2.10.1
[PyYAML](https://pypi.org/project/PyYAML/)==6.0.2
[StrEnum](https://pypi.org/project/StrEnum/)==0.4.15
[cryptography](https://pypi.org/project/cryptography/)==44.0.0
[bcrypt](https://pypi.org/project/bcrypt/)==4.2.1
[jinja2](https://pypi.org/project/jinja2/)==3.1.5
[mohawk](https://pypi.org/project/mohawk/)==1.1.0
[psycopg2](https://pypi.org/project/psycopg2/)==2.9.10
[pyparsing](https://pypi.org/project/pyparsing/)==3.2.1
[pysaml2](https://pypi.org/project/pysaml2/)==7.5.0
[python-dateutil](https://pypi.org/project/python-dateutil/)==2.9.0.post0
[python-ldap](https://pypi.org/project/python-ldap/)==3.4.4
[pytz](https://pypi.org/project/pytz/)==2024.2
[requests](https://pypi.org/project/requests/)==2.32.3
[requests_hawk](https://pypi.org/project/requests_hawk/)==1.2.1
[sentry-sdk](https://pypi.org/project/sentry-sdk/)[[flask]](https://pypi.org/project/[flask]/)==2.19.2
[uWSGI](https://pypi.org/project/uWSGI/)==2.0.28

[alex]

I'm not aware of any change we made that I would expect to pessemize any workload.

Unfortunately, this isn't really enough for us to reproduce/analyze.

At a minimum, we need a minimal reproducer. Ideally, if you could get a profile of the faster/slower versions that we could compare, that'd probably be sufficient for us to diagnose.

[reaperhulk]

We did switch from OpenSSL 3.3.1 to 3.4.0 in this release, but unfortunately we’re going to need a reduced test case here to understand potential regressions.

[lazysteel93]

Thanks, guys. I understand. I'll see if I can figure out a test case.