Releases: irrdnet/irrd
IRRd 4.2.2
IRRd 4.2.2 was released on December 2nd, 2021, and fixes four issues that
occurred in 4.2.1:
- If users created multiple objects with the same RPSL primary key, in the
same source, with different object classes, IRRD would only
import one of them and discarded the other. This could occur, for example,
when amntner
andas-set
had the same name. This issue was quite rare. (#560) - When running on PyPy, IRRD had
infinite file descriptor growth causing it to run out
of file descriptors eventually regardless of configured limits. (#578) - IRRD
was unable to run in the foreground with stdout logging
_. (#557) - IRRD would
reject the autogenerated last-modified attribute
_ in submitted
objects, which was confusing to users. (#587)
Upgrading requires updating the database. See the release notes for the details:
https://irrd.readthedocs.io/en/stable/releases/4.2.2/
IRRd 4.2.1
IRRd 4.2.1 was released on October 29th, 2021, and fixes two issues that occurred in 4.2.0:
- IRRd’s whois worker processes could use excessive memory #571
- When dropping privileges, IRRd did not correctly configure the user’s groups, causing issues with typical Redis setups #567
The memory issue occurred mainly when answering queries with large responses. Those cause high memory consumption in the workers, which is expected, but after answering the query, the workers did not always release the memory.
IRRd 4.1.8
https://irrd.readthedocs.io/en/stable/releases/4.1.8/
IRRd 4.1.8 was released on October 29th, 2021, and fixes two issues that occurred in all previous 4.1 releases:
- IRRd’s whois worker processes could use excessive memory #571
- The scopefilter_enabled value in the !J was incorrect in some cases #555
The memory issue occurred mainly when answering queries with large responses. Those cause high memory consumption in the workers, which is expected, but after answering the query, the workers did not always release the memory.
IRRd 4.2.0
IRRd 4.2.0 release candidate 3
IRRd 4.1.7
IRRd 4.1.7 was released on September 7th, 2021, and fixes an issue where the NRTM parser would accept partial content. In case of poor network connectivity, the TCP connection could break while receiving NRTM data. IRRd would then process the received part, potentially causing updates to be missed.
In 4.1.7, IRRd will reject and log any NRTM response with a valid START line that does not end in %END
IRRd 4.2.0 release candidate 2
https://irrd.readthedocs.io/en/v4.2.0rc2/releases/4.2.0/
Compared to 4.2.0rc1: