release-21.2: security: remove the node TLS client cert exemption

[blathers-crl]

Backport 1/1 commits from #71134 on behalf of @knz.

/cc @cockroachdb/release

---

Fixes #71102.

Release note (security update): It is not possible any more to use a
node TLS certificate to establish a SQL connection with another
username than node. This facility had existed as an "escape hatch"
so that an operator could use the node cert to perform operations on
behalf of another SQL user. However, this facility is not necessary:
an operator with access to a node cert can log in as node directly
and create new credentials for another user anyway. By removing
this facility, we tighten the guarantee that the principal in the TLS
client cert always matches the SQL identity.

---

Release justification: prevents a significant security risk in CC serverless

[blathers-crl]

Thanks for opening a backport.

Please check the backport criteria before merging:

• Patches should only be created for serious issues.
• Patches should not break backwards-compatibility.
• Patches should change as little code as possible.
• Patches should not change on-disk formats or node communication protocols.
• Patches should not add new functionality.
• Patches must not add, edit, or otherwise modify cluster versions; or add version gates.

If some of the basic criteria cannot be satisfied, ensure that the exceptional criteria are satisfied within.

• There is a high priority need for the functionality that cannot wait until the next release and is difficult to address in another way.
• The new functionality is additive-only and only runs for clusters which have specifically “opted in” to it (e.g. by a cluster setting).
• New code is protected by a conditional check that is trivial to verify and ensures that it only runs for opt-in clusters.
• The PM and TL on the team that owns the changed code have signed off that the change obeys the above rules.

Add a brief release justification to the body of your PR to justify this backport.

Some other things to consider:

• What did we do to ensure that a user that doesn’t know & care about this backport, has no idea that it happened?
• Will this work in a cluster of mixed patch versions? Did we test that?
• If a user upgrades a patch version, uses this feature, and then downgrades, what happens?

[cockroach-teamcity]

This change is 

[catj-cockroach]

Reviewed 10 of 10 files at r1, all commit messages.
Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @andy-kimball, @bdarnell, @blathers-crl[bot], @chrisseto, and @knz)

---

pkg/acceptance/cluster/certs.go, line 42 at r1 (raw file):

	maybePanic(security.CreateClientPair(
		certsDir, filepath.Join(certsDir, security.EmbeddedCAKey),
		2048, 48*time.Hour, false, security.RootUserName(), true /* generate pk8 key */))

This looks like an unintentional side effect but one I'm not worried about.

---

pkg/acceptance/compose/gss/psql/start.sh, line 16 at r1 (raw file):

echo "Preparing SQL user ahead of test"
env \
    PGSSLKEY=/certs/client.root.key \

Not certain this is desired... I'll dig into this more.

---

pkg/acceptance/compose/gss/python/start.sh, line 14 at r1 (raw file):

echo psql | kinit [email protected]

export PGSSLKEY=/certs/client.root.key

Not certain this is desired either

[catj-cockroach]

The test failure in Github CI (Cockroach) was for the package ccl/sqlproxyccl/tenant, in the Bazel CI runner the tests passed so I'm rerunning the tests.

[catj-cockroach]

This looks good to me, I don't have any concerns about the code that's being changed by the backport.

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @andy-kimball, @bdarnell, @blathers-crl[bot], @chrisseto, and @knz)

---

pkg/acceptance/compose/gss/psql/start.sh, line 16 at r1 (raw file):

Previously, catj-cockroach (Cat J) wrote…

Not certain this is desired... I'll dig into this more.

Oh I missed that this was for creating a user for testing with, ignore this comment and the following comment regarding PGSSLKEY

[knz]

Reviewable status: complete! 0 of 0 LGTMs obtained (waiting on @andy-kimball, @bdarnell, @blathers-crl[bot], @catj-cockroach, and @chrisseto)

---

pkg/acceptance/cluster/certs.go, line 42 at r1 (raw file):

Previously, catj-cockroach (Cat J) wrote…

This looks like an unintentional side effect but one I'm not worried about.

It's absolutely necessary. The Docker image we use as base for the TestComposeGSS uses a version of debian which has a OpenSSL constraint on minimum key size. At key size 1024, the test fails with error message "ee key too small" when the key is loaded from disk.