-
-
Notifications
You must be signed in to change notification settings - Fork 140
Description
The current security practice for nodejs project is to notify everyone of upcoming security release approximately a week in advance and then do a release centered around these security issues only. The security issues could be Node specific or in the currently bundled 3rd party dependencies.
Security notifications are limited to members of https://groups.google.com/g/nodejs-sec
There are two-fold issues here when it comes to Linux distros (or BSD),
- most distributions are subscribed to issues via oss-security mailing list - https://oss-security.openwall.org/wiki/mailing-lists/oss-security . CC'ing this mailing list would raise awareness of security issues.
- coordinated release dates (CRD) for embargoed bugs via https://oss-security.openwall.org/wiki/mailing-lists/distros
Regarding the second point, and I will use OpenSSL as an example, normally 1 week prior to a public release of a security fix, a member of openssl project would mail the distros mailing list a description of the bug along with pending patches. This allowed distributions to analyze these issues and backport fixes while keeping the issue non-public. Then when the issue is made public (generally by same person that sent the embargoed report in the first place), distributions are ready with the fixes, or at least they've had some time to analyze the impact. There is a header-only archive where you can see this followed few days later by public announcement,
https://www.openwall.com/lists/linux-distros/2020/12/01/8
https://www.openssl.org/news/secadv/20191206.txt
The question that remains is Why? The main answer is that the lifecycle of Node releases in some distributions is longer than the LTS lifecycle in the nodejs project. At SUSE we still have nodejs8 maintained for some customers. Furthermore, the release process needs to account for additional testing by distributions which also requires time. A few days advance warning would prevent this race between the issue being public and being fixable by all the integrators. This was especially important in more visible linux kernel or openssl issues in the recent past.
Finally, this is not about the 3rd party bugs, like bugs in V8 or OpenSSL or ICU (since those issues are public before being fixed in Node), but reserved to the core node code.
So, my questions here are,
- would NodeJS project consider using CRD via the already curated security-centered web-of-trust that is the distros mailing list?
- could oss-security ML be cc'ed when there is a security release from node?