Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 24 Dec 2024 17:27:58 +0000
From:      bugzilla-noreply@freebsd.org
To:        java@FreeBSD.org
Subject:   [Bug 283364] java/openjdk*: Update pkg-message information about fdescfs  and procfs
Message-ID:  <bug-283364-8522-oWW8rSDDci@https.bugs.freebsd.org/bugzilla/>
In-Reply-To: <bug-283364-8522@https.bugs.freebsd.org/bugzilla/>
References:  <bug-283364-8522@https.bugs.freebsd.org/bugzilla/>

next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D283364

Eirik Oeverby <ltning-freebsd@anduin.net> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ltning-freebsd@anduin.net

--- Comment #6 from Eirik Oeverby <ltning-freebsd@anduin.net> ---
We've run java5, openjdk6, 7, 8, 11 and now 17 in production for a couple of
decades now and rarely had procfs or fdescfs mounted. The exception was whe=
n,
for a while, we had to run a Linux JDK because the FreeBSD port wasn't work=
ing
(don't ask me why).

I think the patch (or one like it) makes sense, even before any work has be=
en
done: Suggesting that it is *required* is simply wrong, just like stating t=
hat
it is never needed is wrong. Based on our experience, and after getting Xav=
ier
and friends' help to analyse it more, it seems that the sometimes-quoted
"performance reasons" for having these are dubious at best; it either works=
 or
does not. "There is no try." (ba-dum-tssss)

The potential security impact (which I may or may not be imagining) of havi=
ng
these filesystems mounted in e.g. a locked-down jail running a web applicat=
ion
might be a good enough reason to mention that their presence may be optiona=
l.

(Then again, we may already have spent more time discussing this pkg-message
than any benefit of modifying it might end up having :) )

Examples of not being needed:
- Tomcat/jboss/whatnot (though webapps may or may not require it)
- Most standalone Java applications I've tested

Examples of it being needed (seems to frequently have to do with observing =
the
local system):
- Graylog (otherwise it cannot see if its own inputs are running or not)
- Elasticsearch, Opensearch (varies a bit between versions)
- Logstash
(com.sun.management.internal.OperatingSystemImpl.getCommittedVirtualMemoryS=
ize0)

/Eirik

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-283364-8522-oWW8rSDDci>